Volatiles in protoplanetary disks

Klaus M. Pontoppidan Space Telescope Science Institute

Colette Salyk National Optical Astronomy Observatory

Edwin A. Bergin University of Michigan

Sean Brittain Clemson University

Bernard Marty Universite´ de Lorraine

Olivier Mousis Universite´ de Franche-Comte´

Karin I. Oberg¨ Harvard University

Volatiles are compounds with low sublimation temperatures, and they make up most of the condensible mass in typical planet-forming environments. They consist of relatively small, often hydrogenated, molecules based on the abundant elements carbon, nitrogen and oxygen. Volatiles are central to the process of planet formation, forming the backbone of a rich chemistry that sets the initial conditions for the formation of planetary atmospheres, and act as a solid mass reservoir catalyzing the formation of planets and planetesimals. Since Protostars and Planets V, our understanding of the evolution of volatiles in protoplanetary environments has grown tremendously. This growth has been driven by rapid advances in observations and models of protoplanetary disks, and by a deepening understanding of the cosmochemistry of the solar system. Indeed, it is only in the past few years that representative samples of molecules have been discovered in great abundance throughout protoplanetary disks (CO, H2O, HCN, C2H2, + CO2, HCO ) – enough to begin building a complete budget for the most abundant elements after hydrogen and helium. The spatial distributions of key volatiles are being mapped, snow lines are directly seen and quantified, and distinct chemical regions within protoplanetary disks are being identified, characterized and modeled. Theoretical processes invoked to explain the solar system record are now being observationally constrained in protoplanetary disks, including transport of icy bodies and concentration of bulk condensibles. The balance between chemical reset – processing of inner disk material strong enough to destroy its memory of past chemistry, and inheritance – the chemically gentle accretion of pristine material from the interstellar medium in the outer disk, ultimately determines the final composition of pre-planetary matter. This chapter focuses on making the first steps toward understanding whether the planet formation processes that led to our solar system are universal.

1. Introduction earliest pieces of firm evidence for abundant ices in plan- etary systems came from cometary spectroscopy, showing The study of the role of ices and volatile compounds in the photodissociation products of what could only be com- the formation and evolution of planetary systems has a long mon ices, such as water and ammonia (Whipple 1950). Fol- and venerable history. Up until the late 20th century, due to lowing early suggestions by e.g., Kuiper (1953), the pres- the lack of knowledge of exo-planetary systems, the solar ence of water ice in the rings of Saturn (McCord et al. 1971) system was the only case study. Consequently, some of the

1 and in the Jovian satellite system (Pilcher et al. 1972) was in which parts of the disk are violently reset, while others confirmed using infrared spectroscopy. The characteriza- are inherited and preserved over the lifetime of the central tion of ices in the outer solar system continues to this day star, and where most regions show some evidence for both, (Mumma and Charnley 2011; Brown et al. 2012). Simi- due to a variety of mixing processes. In this chapter, we will larly, the presence of ices in the dense interstellar medium consider the solar nebula as a protoplanetary disk among – the material out of which all planetary systems form – many others – and indeed call it the solar protoplanetary has been recognized since the 1970s and conjectured even disk. We will discuss the issue of inheritance versus reset, earlier. With the advent of powerful infrared satellite ob- and suggest a division of disks into regions characterized by servatories, such as the Infrared Space Observatorory (ISO) these very different chemical circumstances. and the Spitzer Space Telescope, we know that interstellar ices are commonplace and carry a large fraction of the solid 1.2. Defining protoplanetary disks and volatiles mass in protostellar environments (Gibb et al. 2004; Oberg¨ The term protoplanetary disk generally refers to the ro- et al. 2011b). tationally supported, gas-rich accretion disk surrounding The next logical investigative steps include the tracking a young pre-main sequence star. The gas-rich disk per- of the volatiles as they take part in the formation and evolu- sists during planetesimal and giant planet formation, but tion of protoplanetary disks — the intermediate evolution- not necessarily during the final assembly of terrestrial plan- ary stage between the interstellar medium and evolved plan- ets. During the lifetime of a protoplanetary disk, both solid- etary systems during which the planets actually form. This and gas-phase chemistry is active, shaping the initial com- subject, however, has seen little progress until recently. A position of planets, asteroids, comets and Kuiper belt ob- decade ago, volatiles in protoplanetary disks were essen- jects. Many of the chemical properties and architecture tially beyond our observational capabilities and the paths of the current day solar system were set during the nebu- the volatiles take from the initial conditions of a protoplan- lar/protoplanetary disk phase, and are indeed preserved to etary disk until they are incorporated into planets and plan- this day, although somewhat obscured by dynamical mix- etesimals were not well understood. What happened over ing processes during the later debris disk phase. the past decade, and in particular since the conclusion of Most of the mass in protoplanetary disks is in the form PPV, is that our observational knowledge of volatiles in pro- of molecular hydrogen and helium gas. Some of the solid toplanetary disks has been greatly expanded, opening up a mass is carried as dust grains, mostly composed of silicates new era of comparative cosmochemistry. That is, the so- and with some contribution of carbon-dominated material. lar nebula is no longer an isolated case study, but a data This material is generally refractory, that is, very high tem- point – albeit an important one – among hundreds. The peratures (& 1000 K) are needed to sublimate it. Only in emerging complementary study of volatiles in protoplane- a very limited region, in a limited period of time or under tary disks thus feeds on comparisons between the properties unusual circumstances are refractory grains returned to the of current-day solar system material and planet-forming gas gas phase. The opposite of refractory is volatile. In the con- and dust during the critical first few million years of the de- text of this chapter, disk volatiles are molecular or atomic velopment of exo-planetary systems. This is an area of in- species with relatively low sublimation temperatures (< a tense contemporary study, and is the subject of this chapter. few 100 K) that are found in the gas phase throughout a sigificant portion of a typical disk under typical, quiescent 1.1. Emerging questions disk conditions1. It is also possible to define a sub-class of Perhaps one of the most central questions in astronomy volatile material that includes all condensible species – that today is whether our solar system, or any of its charac- is, the class of volatiles that are found in both their solid and teristics, is common or an oddity? We already know that gaseous forms in significant parts of the disk, but which ex- most planetary systems have orbital architectures that do cludes species that never condense in bulk, such as H2. It not resemble the solar system, but is the chemistry of the is the ability of condensible volatiles to relatively easily un- solar system also uncommon? Another matter is the degree dergo dramatic phase changes that lead to them having a to which volatiles are inherited from the parent molecular special role in the evolution of protoplanetary disks and the cloud, or whether their chemistry is reset as part of typi- formation of planets. An example of a volatile that is not cal disk evolution. That is, can we recover evidence for considered condensible is the dominant mass compound, an interstellar origin in protoplanetary and planetary mate- H2, while one of the most important condensible species rial, or is that early history lost in the proverbial furnace of is water. planet formation? There is currently an apparent disconnect between the cosmochemical idea of an ideal condensation 1In cosmochemistry the use of the words refractory and volatile is some- sequence from a fully vaporized and hot early phase in the what different. While they are still related to condensation temperature, their use is usually reserved for atomic elements, rather than molecules, solar nebula, supported by a wealth of data from meteoritic for interpreting elemental abundances in meteorites. material, and the astrophysical idea of a relatively quiescent path of minimally processed material during the formation of protoplanetary disks. The answer is likely a compromise,

2 2. The solar nebula as a volatile-rich protoplanetary observable is illustrated in Figure 1, where the Earth’s rela- disk tive elemental abundances are compared to those of the Sun. While the refractory elements are essentially of solar abun- The solar system was formed from a protoplanetary disk dances, others are depleted by orders of magnitude, most during a time frame of 2-3 million years, spanning ages of prominently carbon and nitrogen, and to a lesser degree, at least 4567-4564 Myr – a number known to a high de- oxygen. Going beyond the Earth, C, N and O abundances gree of precision, thanks to accurate radiometric dating of in various solar system bodies are compared (Earth, mete- primitive meteoritic material formed by gas-condensation orites, comets, and the Sun) in Figure 2. processes (Scott 2007). The composition of solar system bodies, including primitive material in meteorites, comets, and planets provide a detailed window into their formation. 109 Fe Si The primordial elemental composition of the solar proto- 108 Mg O Ni planetary disk has been systematically inferred from mea- 7 Ca Al 10 Cr S surements of the solar atmosphere, generally under assump- Mn 106 Na tions of efficient convective mixing. Any compositional or Ti Co P C

105 Cu isotopic differences measured in primitive solar system ma- V K Zn terials, such as comets and meteorites, can therefore be as- 104 Sc Ge cribed to processes taking place during the formation and Ga 3 Cl

Earth bulk abundance 10 evolution of the solar system, generally after the formation F N of the Sun. 102 One of the initial models to account for the volatility 101 patterns of meteorites is one where the solar disk started 100 102 103 104 105 106 107 108 109 hot (> 1400 K), such that all material was vaporized. As Solar abundance the gas cooled, the elements formed minerals governed by the condensation sequence in thermodynamic equilibrium Fig. 1.— Solid Earth bulk elemental abundances (Allegre` (e.g. Grossman 1972; Wood and Hashimoto 1993; Ebel and et al. 2001) compared to solar abundances (Grevesse et al. Grossman 2000), with some level of incompletion due to 2010). The solar abundances are normalized to an H abun- grain growth and disk gas dispersal (Wasson and Chou dance of 1012. For the Earth, the silicon abundance is scaled 1974; Cassen 1996). Condensation models predict that to that of the Sun. All abundances are in units of relative Calcium-Aluminum compounds would be the first elements number density of nuclei. The dashed line guides the eye to to condense (Lord 1965; Grossman 1972), a proposition the 1:1 ratio. For instances, the solar abundances of carbon, that is consistent with Calcium-Aluminum Inclusions (CAI) oxygen and nitrogen are 108.43, 108.69 and 107.83, respec- being the oldest material found in the solar system (Amelin tively. et al. 2002). The widespread presence of presolar grains, albeit in ex- tremely dilute quantities (Zinner 1998), the high levels of A great amount of information about the relative distri- deuterium enrichments seen in Oort cloud comets and mete- bution of volatiles in the solar nebula is contained in these orites (Mumma and Charnley 2011; Alexander et al. 2012), plots. Specifically, they indicate the volatility of the chemi- and the presence of highly volatile CO in comets, presents cal compounds carrying the bulk of each element: If an el- the contrary perspective that at least part of the nebula re- ement is highly depleted, most of that element must either mained cold or represents material provided to the disk at have been in a form too volatile to condense at the loca- later cooler stages. It is with these fundamental contradic- tion and time that a specific solar system body formed, or tions in mind that we strive to understand the solar system the material was subjected to subsequent, possibly transient, and general protoplanetary disks under a common theoreti- heating at a later time while still in a form where outgassing cal umbrella. and loss could be efficient. In other words, while the plot will not directly tell us what the main elemental carriers 2.1. The bulk abundances of carbon, oxygen and nitro- were, it will provide us with some of their basic physical- gen chemical properties. In the case of the Earth, we can see For instance, consider the bulk abundances of common that many elements, and not only C, N and O, are depleted elements in the Earth, and in particular of carbon, oxy- to some extent, giving rise to the idea that the Earth formed gen and nitrogen, which are the main building blocks, from material exposed to temperatures well above 150 K, along with hydrogen, of volatile chemistry. With a sur- and going as high as 1000 K (Allegre` et al. 2001). face dominated by carbon-based life, coated by water, with The ISM and the Sun represent the total amount of mate- a nitrogen-rich atmosphere, our planet is actually a carbon-, rial available. Beyond the Sun, most bodies in the solar sys- nitrogen- and water-poor world, compared to solar abun- tem are hydrogen-poor. This reflects the fact that molecular dances and, presumably, to the abundance ratios of the hydrogen is the most volatile molecular compound present molecular cloud out of which the solar system formed. This in a protoplanetary disk and is never efficiently incorporated

3 102 O inner disk. From these considerations, likely candidates are O 1 O C H O, CO and CO. 10 C C O 2 2 n C O O o O O The bulk carbon abundance shows a stronger signal c

i N O l 0 i 10 C S C C N C than that of oxygen. As we traverse the solar system, from o t -1 comets over the asteroids to the Earth, we discover a gradi- e 10 N v i

t C N h t a N ent in the relative amounts of condensed carbon, culminat- l

-2 r & e M

10 a S R

y ing in a carbon depletion of three orders of magnitude in the E I N e e l c -3 l & Earth. The caveat is that the Earth’s C/Si ratio is estimated n 10 a N N a n

C f p e H d u r t

p from the mantle (Allegre` et al. 2001), but the Earth’s total : n i a -4 S s u I M V O o 10 h t w b B C C C C e carbon content is highly uncertain as there might be deep A - W D e

-5 m l reservoirs of carbon in the lower mantle and/or core (Wood

10 a o Meteorites N

C H 1993; Dasgupta and Hirschmann 2010). However, mete- 10-6 104 102 100 10-2 10-4 orites have been posited as tracers of the Earth’s starting H/Si materials and even unprocessed carbonaceous chondrites have nearly an order of magnitude carbon deficiency rel- Fig. 2.— The relative CNO abundances in the solar system. ative to the total amount that was available in the nebula. The abundances are computed relative to Si, the primary re- The likely inference is that within the inner nebula volatile fractory element in the solar system and interstellar space. carbon must have been predominantly in a volatile, gaseous The abundances are shown as a function of the H/Si ratio form, in contrast with the ISM, where much, perhaps most, which separates the various bodies. The figure is taken and of the carbon is sequestered as graphite or other highly re- updated from Lee et al. (2010); Geiss (1987). The white fractory forms. This so-called carbon deficit problem will dwarf values are for GD 40 (Jura et al. 2012) and are pro- be discussed further in §2.2. The carbon deficit in planetes- vided with an arbitrary H/Si ratio (since this is unknown). imals is contrasted with a carbon enhancement in the giant planet atmospheres; the C/H in the Jovian atmosphere is enhanced by a factor 3.5 over solar (Niemann et al. 1996; Wong et al. 2004), while the Saturnian ratio is 7 times so- into solids anywhere in the solar system. Comets formed in lar (Flasar et al. 2005). This is consistent with the giant the coldest regions of the nebula and incorporate the largest planets forming outside the condensation front of a primary amount of hydrogen, mostly trapped in the form of water. volatile carbon carrier. In contrast, the rocky planetesimals that formed close to the The carbon that made it to the Earth was likely deliv- Sun contain successively less water, and therefore less hy- ered after the protoplanetary disk phase, but during the bulk drogen. Earth formation, and there is significant evidence that it is The oxygen content in the solar system is also, to a large mostly of chondritic origin (Marty 2012; Albarede` et al. degree, a story of water. Oxygen is generally the least de- 2013). Carbonaceous chondrites are composed primarily of pleted element among the CNO trifecta across the solar sys- silicates, but contain up to 10-20% equivalent water in the tem, indicating that a relatively more refractory compound form of hydrated silicates, and a few percent of carbon, by carried much of the oxygen. A likely candidate for this car- mass (Scott 2007). While strongly depleted in bulk carbon rier is silicates (mostly oxides built with the anions SiO4− 4 relative to solar abundances, the remaining chondritic car- or SiO2−). Indeed, comets have the highest O/Si ratio, 3 bon (and nitrogen) is mostly carried by refractory organics. above that even of the Sun, although there is some uncer- The origin of chondritic organics is often traced by the iso- tainty in the calculation of silicon abundances in comets as topic ratios of noble gases (Ar, Ne, Xe, ...), trapped in an these depend on an assumed mass opacity (Min et al. 2005). ill-defined organic phase (called “Q”; Ott et al. 1981; Mar- However, one value comes from in situ measurements of rocchi et al. 2005). Present in very small fractions (< 1% Comet Halley by the Soviet Vega-1 spacecraft (Jessberger by mass), chondritic noble gases generally have a recurring et al. 1988), and supports this general picture. Thus, comets isotopic signature, strongly fractionated relative to that of incorporated oxygen in the form of silicates, as well as the Sun and the solar protoplanetary disk (chondritic noble CNO ices. Rocky bodies also contain significant amounts gases are heavy). The isotopic signature in noble gases in of oxygen, but below the total atomic nebular reservoir, by the Earth’s atmosphere is consistent with a mixture that is factors of 2-10. This allows a rough estimate of the frac- 90% chondritic and at most 10% solar (Marty 2012). The tion of oxygen in volatiles versus refractory compounds – a chondritic noble gas fractionation has been reproduced in few volatile oxygen atoms for each refractory oxygen atom. the laboratory during gas-solid exchanges under ionizing Furthermore, it demonstrates that in the terrestrial planet conditions (Frick et al. 1979), consistent with entrapment forming regions of the nebula (inside the snow line), most during photochemical formation of the organics from ices oxygen was in a gaseous form. Overall, the variation in in the solar disk (Jenniskens et al. 1993; Gerakines et al. bulk oxygen abundance throughout the solar system sug- 1996; Oberg¨ et al. 2009). Another way of correctly frac- gests that the main oxygen carrier was a condensible species tionating noble gases is by selective trapping in forming — frozen out in the outer disk, yet in the gas-phase in the ices. Notesco et al. (1999) found experimentally that noble

4 gases are enriched according to their isotopic mass ratios terstellar medium sequesters ∼ 60% of the carbon in some p ( m1/m2) during the formation of amorphous water ice. unknown combination of amorphous carbon grains, large That is, a similar effect can be generated as a purely thermal organics, or, perhaps, as polycyclic aromatic hydrocarbons effect. In summary, there is significant evidence that the (PAHs) (Savage and Sembach 1996; Draine 2003). If ter- Earth’s carbon is delivered with a volatile reservoir. That restrial worlds formed directly from solid grains that are said, it is not clear whether the delivery vector was from supplied to the young disk via collapse then they would a chondritic reservoir, as opposed to a cometary reservoir. have 2 times more carbon than silicon by mass. This is There is support for chondritic delivery, but an unambigu- strikingly different from the observed ratio of 1-2% of C/Si ous identification was for a long time not possible due to a by mass (Allegre` et al. 2001) and essentially all of the lack of accurate measurements of noble gases in comets. primordial, highly refractory, carbon carriers must be de- Recently, samples returned by the stardust mission from stroyed prior to terrestrial planet formation to account for comet 81P/Wild 2 showed similar noble gas signatures as the composition of the Earth and carbonaceous chondrites. those of chondrites, indicating that the noble gas hosts may A similar signature is known to exist in exoplanetary be of similar origin in comets and chondrites (Marty et al. systems from measurements of asteroidal material accreted 2008). onto white dwarfs (Jura 2006). A small fraction of white Nitrogen is the most depleted element in our selection. dwarfs show lines from a wide range of heavy elements. In the Earth, it is depleted by 5 orders of magnitude. While Since any heavy elements are rapidly depleted from white the Earth’s atmosphere is dominated by N2, this represents dwarf photospheres by settling, their presence indicates about half of the Earth’s nitrogen (Marty 1995). In com- that they must have been recently accreted (Paquette et al. parison, the total mass of the atmosphere is a bit less than 1 1986). In such polluted white dwarfs, the C/Fe and C/Si ppm of the total mass of the Earth. While the absolute abun- abundance ratios are well below solar and consistent with dance of nitrogen is uncertain in comets, it is clear that it is accretion of asteroidal material deficient in carbon (Jura highly depleted relative to the Solar value. The best esti- 2008; Gansicke¨ et al. 2012; Farihi et al. 2013). This sug- mates come from the Halley in situ measurements (Wyckoff gests that the Earth’s missing carbon and the carbon deficit et al. 1991) as this accounts for N incorporated in the coma in the inner solar system are a natural outcome of star and as evaporated ices and in the refractory component. The planet formation, that is, the process leading to a strong car- Hale-Bopp estimate (taken from the summary of Mumma bon deficit is a universal property of protoplanetary disks. and Charnley 2011), which has the lower N/Si ratio in Fig- Theoretical models suggest the carbon grains can be de- ure 2, only includes the ices and is therefore a lower limit. stroyed via oxidation either by OH at the inner edge of the One interesting aspect is that Wyckoff et al. (1991) estimates accretion disk (Gail 2002) or by oxygen atoms created by that Comet Halley carried most of its nitrogen (90%) in photodissociation of oxygen-carrying volatiles, such as wa- a carbonaceous (CHON) refractory component. As in the ter, on disk surfaces exposed to ultraviolet radiation from other heavy element pools, there is a gradient in the N/Si the central star (or other nearby stars) (Lee et al. 2010). Al- ratio as we approach the terrestrial planet forming region though the details vary depending upon the uncertain car- with both meteorites and the Earth exhibiting large nitrogen rier of carbon grain absorption in the interstellar medium, depletions, relative to the total amount available. Clearly, the destruction of the grains could produce CO or CO2 the vast majority of the nitrogen resided in the gas in the (graphite/amorphous carbon) or, to a lesser degree, hydro- solar nebula in a highly volatile form, which could be NH3, carbons with oxygen from PAHs (Gail 2002; Lee et al. N2 and/or N. In this context, laboratory measurements sug- 2010; Kress et al. 2010). The theoretical implication is that gest that both CO and N2 have similar binding strengths to CO or CO2 are the likely dominant carriers of carbon in the the grain surface (Oberg¨ et al. 2005). Thus the difference inner disk, rather than either solid-phase carbon grains or between the presence of CO in comets and the relative lack gas-phase molecules like CH4, and that the sum of these of N2 stands out. Note, however, that in the Jovian atmo- species should be close to the solar carbon abundance. sphere, nitrogen in the form of NH3 is enhanced relative to solar by a factor 3 (Wong et al. 2004), similar to that of 2.3. The solar system snow line carbon, and similarly indicating that the Jovian core formed We know that the distribution of planets in the solar sys- outside of the primary nitrogen condensation front. tem is skewed, with low-mass terrestrial planets residing inside of 3 AU and massive gas and ice giant planets or- 2.2. Origin of the carbon deficit biting beyond this radius, and it has long been thought that As we have seen in Figure 2, solids in the terrestrial re- this is related to the presence of a water ice condensation gion of the solar system, as represented by the Earth and front (the so-called “snow line”) being located at a few AU meteorites, have far less elemental carbon than what was (Hayashi 1981). However, the present-day distribution of available, on average, in the solar protoplanetary disk. The water in the asteroid belt is somewhat ambiguous, and the implication is that the carbon was either entirely missing location of the end-stage snow line in the solar disk may in the terrestrial region, or it was in a volatile form, not be preserved only as a fossil. The lifetime of water ice on available for condensation. In stark contrast, the dense in- the surfaces of all but the largest asteroids is short-lived in-

5 side of 5 AU (Lebofsky 1980; Levison and Duncan 1997). particular relative to the many rare refractory systems tra- Exceptions are 1 Ceres, which has some amount of surface ditionally used in cosmochemistry, they are also accessible, water ice (Lebofsky et al. 1981), and 24 Themis (Campins albeit with difficulty, to astronomical observations of the in- et al. 2010; Rivkin and Emery 2010). In both cases, this may terstellar medium and protoplanetary disks. These elements be evidence for subsurface ice revealed by recent activity. show great isotopic heterogeneity in various solar system For most asteroids, the water content has been modified by reservoirs, in sharp contrast to more refractory elements, of heating within a few Myr of disk formation. This heating which most are isotopically homogenized at the 0.1% level led to the melting of interior ice, followed by aqueous alter- — including elements with somewhat lower condensation ation and the formation of hydrated silicates that are stable temperatures, such as Cu, K and Zn (Zhu et al. 2001; Luck up to the current age of the solar system. et al. 2001). Of particular interest is meteoritic oxygen, The direct methods searching for asteroid surface wa- which scatters along the famous mass-independent fraction- ter in the form of hydrated silicates have encountered some ation line in a three-isotope plot (Clayton et al. 1973; see interpretational difficulties. Hydrated silicates on the sur- Figure 3). The 14N/15N ratio exhibits large variations of faces of outer belt asteroids with semi-major axes in the up to a factor 5 among different solar system bodies (Bri- range 2.5- 5.2 AU do not show clear evidence for a snow ani et al. 2009). The ubiquitous strong fractionation of the line (Jones et al. 1990), and their presence may therefore CNOH systems suggests that these elements were carried be more a tracer of local thermal evolution than the pri- by molecular compounds accessible to gas-phase fractiona- mordial water ice distribution (Rivkin et al. 2002). That is, tion processes. they may be a tracer of planetesimals that were exposed to We now know the location of the primordial material transient heating strong enough to melt water and cause the in fractionation diagrams. Recently, the Genesis mission formation of the hydrated silicates, rather than the presence accurately measured the elemental composition of the Sun of water per se. and, by extension, that of the solar protoplanetary disk (Bur- The localization of the solar system snow line relies on nett et al. 2003). Over a period of three years, the solar wind the measured water content of meteorites, coupled with was sampled by passive implantation of ions into pure target spectroscopic links to taxonomic asteroid classes (Chap- material. The spacecraft was located at Lagrangian point L1 man et al. 1975; Chapman 1996). Carbonaceous chondrites to prevent any alteration of the solar wind composition by (CI and CM) contain up to 5-10% of water, by mass (Ker- the terrestrial magnetic field. A result from Genesis of par- ridge 1985), and have been associated with C-type aster- ticular interest to this chapter was the measurement of the oids, which have semi-major axes predominantly beyond isotopic composition of solar oxygen, nitrogen and noble 2.5 AU (Bus and Binzel 2002). Conversely, ordinary chon- gases. Because the young Sun burned deuterium, the D/H drites are relatively dry (0.1% of water, by mass) (Mc- ratio of the solar disk has been measured independently us- Naughton et al. 1981) and are associated with S-type aster- ing the Jovian atmosphere. Genesis established that all solar oids, which are found within 2.5 AU (Bus and Binzel 2002). system reservoirs (except for the atmosphere of Jupiter and The different asteroid types, however, are radially mixed, probably of the other giant planets) are enriched, relative to and any present-day water distribution is not as sharp as the the sun and solar protoplanetary disk, in the rare and heavy snow line once was. This is likely, at least in part, related isotopes of hydrogen (D), nitrogen (15N), oxygen (17O and to dynamical mixing of planetesimals as described by the 18O). There are also indications of an enrichment in 13C in Nice and Grand Tack models (Walsh et al. 2011). solar system solids and gases relative to the Sun (Hashizume In summary, the study of the solar nebula snow line is as- et al. 2004). sociated with significant uncertainties due to a lack of in situ The origin of the fractionation patterns in CNOH in the measurements of asteroids and smearing out by dynamical solar system is still a matter of vigorous debate, and, while mixing that astronomical measurements of snow lines in data have improved in the past decade, few questions have protoplanetary disks may provide important new constraints been conclusively settled; we still do not have a clear idea of on this critical feature of a forming planetary system. whether isotopic ratios in volatiles have an interstellar ori- gin or whether they evolved as a consequence of protoplan- 2.4. Isotope fractionation in solar system volatiles etary disk evolution. In meteorites, the organic matter tends One of the classical methods for investigating the origin to show the strong isotopic anomalies of hydrogen and ni- of solar system materials is the measurement and compari- trogen (Epstein et al. 1987). In the interstellar medium the son of accurate isotopic ratios. This is in particular true for 14N/15N ratio is typically just above 400 with relatively lit- CNOH as the main volatile building blocks, and strong frac- tle variation (Aleon´ 2010, and references therein). This is tionation patterns are indeed observed in the bulk carriers matched by the ratio in the Jovian atmosphere measured of these elements in the solar system. The most important by the Galileo probe (435±57, Owen et al. 2001), and in systems include the deuterium/hydrogen (D/H) fraction, the the Sun (442±66, Marty et al. 2010), but contrasts with relative amounts of the three stable isotopes of oxygen, 16O, cometary and meteoritic values of <150 (Aleon´ 2010, and 17O and 18O, as well as the ratio of the two stable nitrogen references therein). The D/H ratio in organics is similarly isotopes, 14N/15N. Since these are abundant species – in enhanced by factors of up to five in highly localized and heterogenous “hot spots” in meteorites, which are likely ex-

6 plained by the presence of organic radicals with D/H as high as ∼0.02 (Remusat et al. 2009). The D/H ratio is generally correlated with the 14N/15N fractionation in organics in me- teorites and comets, relative to planetary atmospheres and the Sun (see Figure 4). Given the high levels of deuteration observed in the interstellar medium (Ceccarelli et al. 2007) and in protoplanetary disks (Qi et al. 2008), it is tempting to use the meteoritic patterns to argue for an interstellar ori- gin. However, the deuteration in dense clouds and disks is typically higher than in meteorites, leading to disk models that use a highly fractionated initial condition that evolves toward a more equilibrated (less fractionated) system at the high temperatures and densities in a disk (e.g. Drouart et al. 1999). In this way, the protoplanetary disk phase may rep- resent a modified “intermediate” fractionation state, rela- tively close to the solar D/H ratio, compared to the extreme Fig. 3.— The three-isotope plot of oxygen in the solar primordial values. This is discussed in greater detail in the system, showing the fractionation line of meteorites along context of comets in §2.5. We also refer the reader to the a mass-independent slope-1 line. The solar point is a re- chapter in this volume focusing specfically on deuterium cent addition from Genesis samples of the solar wind. This fractionation (Ceccarelli et al., this book). point, labeled “Sun”, is deduced from the sample measure- Oxygen isotopes in the solar system uniquely display a ments, labeled “SW”, by assuming a correction for mass- puzzling mass-independent fractionation pattern (Clayton dependent fractionation in the solar wind. The inset figure et al. 1973), as shown in detail in Figure 3. The location shows an extended view highlighting the very 16O depleted of nearly all inner solar system materials on a single mass- samples of heavy water in the carbonaceous chondrite Acfer independent fractionation line led to a search for physical 094. Figure originally from McKeegan et al. (2011). processes capable of producing the observed fractionation slope, and affecting a large portion of the inner solar neb- ula. A leading candidate is isotope-selective photodissoci- ation of CO (Kitamura and Shimizu 1983; Thiemens and cesses in some grains (Sandford et al. 2006), consistent with Heidenreich 1983; Clayton 2002; Yurimoto and Kuramoto unprocessed materials in the outer solar system, where the 2004; Lyons and Young 2005), a process that can now be comet is believed to have formed. However, the cometary investigated in protoplanetary disks, as we will discuss in samples seem to most resemble samples of primitive mete- Section 3.7. orites, believed to have formed much closer to the young sun. The oxygen, deuterium and nitrogen isotopic com- 2.5. The origin of cometary volatiles in the solar sys- positions of most Wild 2 grains are strikingly similar to tem carbonaceous chondrite meteorites (McKeegan et al. 2006), In the recent review of cometary chemistry by Mumma and neon isotope ratios of noble gases are similar to those and Charnley (2011), the observed composition of comets trapped in chondritic organics (Marty et al. 2008). In addi- is compared to that of the interstellar medium – due to a lack tion, the grains are primarily crystalline, rather than amor- of appropriate observations of protoplanetary disks. Using phous, with compositions inconsistent with being formed recent disk data, we can now begin to make a more direct via low-temperature condensation of interstellar grains, and comparison. The motivation for moving beyond the solar therefore requiring a high temperature origin (Brownlee system is clear: Comets sample some of the oldest and most et al. 2006). Thus, Stardust suggests that cometary matter primitive material in the solar system, and are likely essen- is a mixture of outer and inner solar system materials. tially unaltered since the protoplanetary disk stage. They An alternative tracer of formation location may be the are thus our best window into the volatile composition of spin temperature of water. In molecules with two or more the solar protoplanetary disk. A first, essential question to hydrogen atoms, transitions that change the total nuclear ask in preparation for a comparison to contemporary pro- spin are strictly forbidden. For water, therefore, as long toplanetary disks is: Is the volatile reservoir sampled by as the chemical bonds are not broken, it is expected that comets formed in the solar protoplanetary disk, or is it, in- the ratio between the ortho (s=0) and para (s=1) states per- stead, interstellar? sists at the value set at the formation temperature (spin The analysis of grains from the Stardust mission sug- temperature) of the molecule. Thus, the ortho/para ratio gests that the answer is both, with captured samples of in comets could potentially measure the formation location comet Wild 2 representing instead a wide range of origins. of the comet, and determine the relative contributions of in The comet has incorporated some presolar material (Brown- situ ice formation vs. inheritance of ice from the molecular lee et al. 2006), as evidenced by deuterium and 15N ex- cloud. In comets, the water spin temperature has been mea- sured to span from 26 K (Bockelee-Morvan´ et al. 2009) to

7 more than 50 K (Mumma et al. 1988). These temperatures are significantly higher than those of the water-forming re- gions in molecular clouds (10-15 K), and perhaps represent a wide range of formation radii — 5–50 AU for typical pro- toplanetary disks (D’Alessio et al. 1997; Chiang and Gol- dreich 1997). However, recent laboratory results compli- cate this picture considerably; two experiments show that the spin temperature of water produced at very low temper- ature and subsequently sublimated retains no history of its formation condition (Sliter et al. 2011; Hama et al. 2011). Deuterium fractionation is another tracer of the cometary formation environment. The water ice in Oort cloud comets are typically enriched by a factor of two in deuterium (nor- malized to 1H) with respect to the terrestrial value and most primitive meteorites. However, the recent analysis of Jupiter comet 103P/Hartley2 (Hartogh et al. 2011) and Fig. 4.— Relation between the nitrogen (15N/14N) and deu- Oort comet Garradd (Bockelee-Morvan´ et al. 2012) have terium (D/H) fractionation in the solar system. Figure from shown that lower, even terrestrial-like, D/H ratios exist in Marty (2012). comets. These observations were used to support the idea that Earth’s water was delivered by comets – a hypothe- sis that was otherwise falling out of favor based on the A final complication in the story of comets is that it is high D/H ratios in Oort comets. However, it also suggests now understood that the orbital parameters of a given comet that there is heterogeneity in the formation environments of are a poor tracer of its formation radius in the solar disk. comets, and that there may even be a gradient in the water The Nice model (Gomes et al. 2005) suggests that comets D/H ratios in the solar disk, as predicted by models of the forming anywhere from 5 to >30 AU may have contributed solar nebula (Horner et al. 2007). This suggests a partial to both the Kuiper Belt comets as well as the Oort cloud disk origin for cometary volatiles, or at least that a uniform comets. Cometary origins may be even more interesting if, interstellar D/H ratio evolved during the protoplanetary disk as suggested by Levison et al. (2010), ∼50% of comets are phase. inherited from other stellar systems. The basic idea underlying this model is that the solar disk starts with an initial condition of highly deuterated wa- 2.6. The total mass of condensible volatile molecules ter, inherited from the protosolar molecular cloud. The wa- ter then evolves, via deuterium exchange with HD, toward The magnitude of the local solid surface density is linked less deuteration during the disk phase – a process that hap- to the efficiency of planetesimal formation, and by exten- pens faster at high temperatures in the inner disk, but not sion the formation of planets via accretionary processes. fast enough in the outer disk to account for the observed Hayashi (1981) originally used an ice/rock ratio of 4 in D/H ratios throughout the solar system (in chondrites as his formulation of the classical minimum-mass solar nebula well as comets; Drouart et al. 1999). Consequently, models (MMSN). Some investigations in later years have revised have invoked various mixing processes, such as turbulent this value slightly downwards, but it is still high. Steven- diffusion or viscous expansion, to reproduce the observed son (1985) found a value of 2–3, based on comparisons to deuteration from in-situ disk chemistry (Mousis et al. 2000; the icy moons of the solar system giant planets. Lodders Hersant et al. 2001; Willacy and Woods 2009; Jacquet and (2003) found ice/rock = 2, based on equilibrium chemistry. Robert 2013). In particular, Yang et al. (2012) find that The recent update to the MMSN by Desch (2007) also uses Jupiter-family comets can obtain a terrestrial D/H ratio in ice/rock = 2, while Dodson-Robinson et al. (2009) used a the outermost part of a viscously-evolving disk. The final full chemical model coupled with a dynamical disk model solar system D/H ratios are therefore likely dependent on to predict an ice/rock ratio close to the original ice/rock=4. both inheritance and in situ chemical evolution. This variation is in part due to differences in assumptions The cometary 14N/15N ratios have been measured in on the elemental oxygen abundance, and on the chemistry. both CN and HCN. In these species, the nitrogen isotopic The higher values for the ice/rock ratio require that both ratio is a factor of 2 lower than that of the Earth and most the bulk carbon and nitrogen are readily condensible with meteorites, and a factor of 3 lower than that of the Sun. sublimation temperatures higher than those of e.g., CO and Enrichment of the heavy 15N isotope has been proposed to N2. Typically, this requires carriers such as CH4 and NH3. be evidence for interstellar fractionation, although chemical A high value for the ice/rock ratio therefore predicts large models of dense clouds do not readily produce fractionation amount of gaseous CH4 and NH3 just inside their respective in any nitrogen carrier as strong as that seen in the solar snow lines. The theoretical expectation for protoplanetary system (Terzieva and Herbst 2000), and the usability of the disks is therefore that the value most likely lies somewhere nitrogen isotopic ratio as an origin tracer remains unclear. between ice/rock∼2–4, and precise observations of all the

8 major carriers of carbon, oxygen and nitrogen are needed to determine the true ice/rock ratios, as well as their potential Mass distribution in ISM ice and dust 91% Oxygen variation from disk to disk. 3.0 XCN In the dense interstellar medium and in protostellar en- CH3 OH velopes, a nearly complete inventory of ice and dust has NH3 76% Oxygen 2.5 CO2 CH4 been observed and quantified (e.g., Gibb et al. 2004; Whit- 71% Oxygen ¨ tet et al. 2009; Oberg et al. 2011b) using infrared absorption 4 CO ice O i spectroscopy toward young stellar objects and background S 2.0

o t stars. Whittet (2010) finds that as much as half of the oxy- CO gas e v i gen is unaccounted for in dense clouds when taking into t a

l 1.5 e r account refractories – such as silicates, as well as volatiles s H2 O s in the form of water, CO2 ices and CO gas. This poten- a tially lowers the ice/rock ratio if the missing oxygen is in M 1.0 the form of a chondritic-like refractory organic phase (as suggested by Whittet) or as extremely volatile O2. In the 0.5 SiO4 densest regions of dark clouds and protostellar envelopes, it has been suggested that the ice abundances increase, in 0.0 particular those of H2O, CO2 and CH3OH, and some ob- ICM YSO Extreme YSO servations find that as much as 90% of the oxygen may be accounted for in one region (Pontoppidan et al. 2004; see also Figure 5)). In the case where the ice abundances are Fig. 5.— The mass distribution of molecules in the in- the highest, this yields an observed ice/rock ratio of at least terstellar medium derived from ice observations (Whittet ∼ 1.5. Strictly speaking, this is a lower limit, since some et al. 2007; Boogert et al. 2008; Oberg¨ et al. 2011b). The ice species are as yet unidentified (Boogert et al. 2008), and distribution assumes that all available silicon is bound in some species, such as N2 (Pontoppidan et al. 2003), may the form of tetrahedral silicates, such as forsterite (taking not be detectable. Importantly, in the ISM, NH3 is not very up four oxygen atoms per silicon atom), although other abundant (∼ 5% relative to water, Boogert et al. 2008). The common silicate species contain slightly less oxygen (three remaining uncertain components are the main reservoirs of oxygen atoms per silicon atom), including enstatite and carbon and nitrogen, with detected reservoirs of the former forsterite. The plot displays only observed carriers (ex- only accounting for 35% of the solar abundance, and de- cept for silicates) for three environments: The intra-cloud tected reservoirs of the latter accounting for less than 10% medium (ICM), young stellar objects and the “extreme” of the solar abundance. Likely reservoirs in the ISM are YSO with the highest observed ice abundances from Pon- refractory carbon and N2, neither of which is directly iden- toppidan et al. (2004). Above each column is shown the tifiable using infrared spectroscopy. The difference up to fraction of the solar elemental oxygen abundance that is in- the putative solar value of ice/rock = 4 is essentially recov- cluded in each budget. The observed oxygen fractions are ered if the missing oxygen, carbon and nitrogen are fully slightly higher than in Whittet (2010) due to our use of the sequestered into ices; in Dodson-Robinson et al. (2009) for lower Grevesse et al. (2010) value for O/H = 4.89 × 10−4, −4 example, 50% of the C is in the form of CH4, and 80% of instead of the O/H = 5.75 × 10 value of Przybilla et al. N in the form of NH3. Thus, whether the bulk disk chem- (2008). istry resembles the ISM or not is of fundamental importance for computing the disk’s solid surface density. Recent ob- although of course any disk has a full three-dimensional servations have begun to reveal some of the chemical de- structure. Radially, disks are often divided into vaguely mographics of protoplanetary disks, as we shall see in the defined regions that are roughly separated by an assumed following section. mode of planet formation: Terrestrial (0-5 AU), giant planet 3. The distribution and evolution of volatiles in proto- (5-20 AU) and Kuiper/cometary belt (>20 AU). The first planetary disks two are characterized by a series of condensation fronts, and are often bundled into an “inner disk” nomenclature. The Protoplanetary disks are structures characterized by a latter region is universally known as the “outer disk”. Verti- wide range of physical environments. They contain regions cally, disks are thought of as consisting of three regions — a with densities in excess of 1015 cm−3 and ISM-like condi- low-density, molecule-poor region near the disk surface, an tions with densities < 104 cm−3, they have temperatures intermediate region, with sufficient radiation shielding and in the range 10 K < T < 10,000 K, and parts of them are temperatures to sustain and excite gas-phase molecules (of- deeply hidden beneath optically thick layers of dust with ten called the “warm molecular layer”), and a colder region depths of more than 1,000 magnitudes. From an astro- near the disk midplane which is often shielded by optically physical perspective, disk conditions are often character- thick dust and in which volatile molecules may condense ized in radial or vertical terms, depending on circumstances, beyond a certain radius. A more general review of pro-

9 toplanetary disk structure as it relates to chemistry can be led to ambiguities between ice located in the disk itself and found in Henning and Semenov (2013). ice located in foreground clouds (Pontoppidan et al. 2005). The methods used to infer the presence and relative More recently, sensitive searches for ices in edge-on disks abundances of disk volatiles draw from a heterogenous set have made progress in detecting ice absorption in edge-on of observational facilities, spanning the ultraviolet to ra- disks with no potential contribution from cold foreground dio regimes. Advances in observational capabilities across cloud material (Aikawa et al. 2012). In addition, McClure the electromagnetic spectrum have allowed more compre- et al. (2012) find a tentative detection of water ice emission hensive descriptions of the bulk chemical composition of from a disk observed with the Herschel Space Observatory, protoplanetary disks than previously possible. In place of and recent upgrades to PACS data reduction routines may a picture in which disks are composed simplistically of enable additional detections in the near future. molecular hydrogen and helium with trace amounts of re- Exploration of the molecular content of warm inner disks fractory dust, they are now seen as complex environments (. 10 AU) has greatly expanded since PPV. Inner disks with distinct regions characterized by fundamentally differ- have been readily detected in CO ro-vibrational lines (Na- ent chemical compositions. Some of these regions are more jita et al. 2003; Blake and Boogert 2004), where it is es- accessible to observations than others, and the more read- sentially ubiquitous in protoplanetary disks (Brown et al. ily observed regions are often used to infer the properties 2013). Carr et al. (2004) detected lines from hot water near of unseen or hidden regions. With a few notable exceptions 2.5 µm toward one young star and argued that the origin was of disks with very high accretion rates, protoplanetary disks in a Keplerian disk. However, it was with the launch of the are generally characterized by having a surface that is hotter Spitzer Space Telescope in August 2003 that this field un- than the interior. This property leads to observables (line as derwent a major advance in understanding. C2H2, HCN and well as continuum emission) that are dominated by the sur- CO2 were first detected by Spitzer in absorption toward a face of the molecular layer. This tends to be true even at young low-mass star (Lahuis et al. 2006), with C2H2, HCN radii and at observing wavelengths where the disks are op- absorption also being detected from the ground in a sec- tically thin in the vertical direction (Semenov et al. 2008). ond source (Gibb et al. 2007). But Spitzer also initiated the Thus, most of our direct knowledge of volatiles in disks per- very first comprehensive surveys of the molecular content tains to surface conditions and to a fraction of the vertical in the inner regions of protoplanetary disks. Carr and Na- column density. jita (2008) realized that the mid-infrared spectrum from 10- 20 µm of the classical T Tauri star AA Tau showed a low- 3.1. History of observations of volatiles in disks level complex forest of molecular emission, mostly from pure rotational lines of water, along with lines and bands Until a few years ago, direct measurements of the from OH, HCN, C2H2 and CO2. The emission was charac- amount and distribution of bulk volatiles in protoplane- teristic of abundant water with temperatures of 500-1000 K. tary disks were limited. Part of the reason was that the Salyk et al. (2008) discovered an additional two disks with strong transitions of the most common volatile species were strong mid-infrared molecular emission, and showed that not visible from the ground and space-based observatories these disks also have emission from hot water as traced by were still too insensitive to obtain spectroscopy of typical ro-vibrational transitions near 3 µm. A survey of more than protoplanetary disks. Millimeter observations of a small 60 protoplanetary disks around low-mass (. 1 M ) stars number of large, massive disks revealed the presence of revealed the presence of mid-infrared emission bands from trace species (. 1% of the total volatile mass) in the cold, organics (HCN in up to 30% of the disks, and up to 10% + outer regions of disks (beyond 100 AU), such as CS, HCO for C2H2)(Pascucci et al. 2009). Advances in the data re- and HCN (Dutrey et al. 1997; Thi et al. 2004; Oberg¨ et al. duction procedure for high resolution Spitzer spectra subse- 2010). An exception to these millimeter-detected trace quently revealed water emission, along with emission from species is CO, which is one of the most important bulk the organics, OH, and CO2, in about half the sources in volatiles. A key conclusion of the early millimeter work a sample of ∼ 50 disks around low-mass stars (Pontopp- was that efficient freeze-out leads to strong depletions in idan et al. 2010a). Following the initial flurry of results the outer disk (e.g., Qi et al. 2004). This was supported by based on Spitzer data, the later years have focused on mod- a few detections of ices in disks. Malfait et al. (1998) and eling and interpreting these data (Salyk et al. 2011; Carr Chiang et al. (2001) discovered emission from crystalline and Najita 2011), as well as obtaining follow-up observa- water ice in two disks in the far-infrared. At shorter wave- tions in the near- and mid-infrared of e.g., water and OH, lengths, ices do not produce emission features since the by ground-based facilities at high spectral resolution (Pon- temperatures required to excite existing resonances below toppidan et al. 2010b; Mandell et al. 2012). ∼40 µm will effectively desorb all bulk volatile species, including water. Scattered near-infrared light has since re- 3.2. Retrieval of volatile abundances from protoplane- vealed water ice in at least one disk (Honda et al. 2009), tary disks and attempts have been made to measure ices in absorption toward edge-on disks (Thi et al. 2002), although difficulties Before proceeding to the implications of the observa- in locating the absorbing material along the line of sight has tions of volatiles in protoplanetary disks, it is worthwhile

10 Carr & Najita Salyk et al.

to briefly consider the technical difficulties in accurately Oxygen Carbon Nitrogen measuring the basic physical and chemical parameters of disks from observations of molecular emission lines. The 1.0 assumptions that are imposed on the data analysis essen- tially always generate model dependencies that can cloud SiO4 interpretation. For this reason intense work has been un- 0.8 derway for the past decade to minimize the model depen- dencies in this process and to better understand the physics H2 O that forms emission lines from disks, as well as other anal- 0.6 ogous nebular objects. Retrieval refers to the modeling process of inverting line fluxes and profiles to fundamen- 0.4 Carr & Najita tal physical parameters such as the fractional abundance (X(species) = n(species)/n(H)(R, z)), the total gas mass CO Salyk et al. or kinetic gas temperature. 0.2 CO Because molecular emission line strengths are depen- dent on abundance as well as excitation, retrieval is neces- sarily a model-dependent, usually iterative, process, even 0.040 for disks that are spatially resolved. A common proce- 0.035 dure uses broad-band multi-wavelength observations (1- Fraction of element 1000 µm) to measure the emission from dust in the disk. Structural models for the distribution of dust in the disk 0.030 are fit to the data, typically under some assumption of dust 0.025 properties. Assuming that the dust traces the bulk molecu- OH lar gas, the spatial fractional abundance structure of molec- 0.020 ular gas-phase species can be retrieved by modeling the line emission implied by the physical structure fixed by the dust 0.015 (Zhang et al. 2013; Bergin et al. 2013). Another analysis approach does not attempt to retrieve molecular abundances 0.010 CO2 CO directly, but attempts to constrain the global disk physical 2 structure by fixing the input elemental abundances and us- 0.005 HCN ing a thermo-chemical model to calculate line fluxes. Such C H 0.000 2 2 HCN models simultaneously solve the chemistry and detailed ra- Oxygen Carbon Nitrogen diative transfer, but rely on having incorporated all impor- tant physical and chemical processes (e.g., Aikawa et al. Fig. 6.— Observed elemental abundance budget at ∼1 AU 2002; Thi et al. 2010; Woitke et al. 2011; Tilling et al. 2012). in protoplanetary disks. The values retrieved by Salyk et al. Both approaches carry with them significant uncertainties (2011), Carr and Najita (2008) and Carr and Najita (2011) and degeneracies, so molecular abundances from disks are are compared side-by-side to provide a rough indication of probably still inaccurate to an order of magnitude, although current uncertainties. All budgets are normalized to the progress is continuously being made to reduce this uncer- assumption that all oxygen is accounted for in the warm tainty (Kamp et al. 2013). molecular layer (fraction of O= 1). With this assumption, about a 3rd of the carbon is accounted for, and almost none 3.3. Demographics of volatiles in disks of the nitrogen. The implication is that any missing con- It is now possible to summarize some general observed tribution is located in an as-yet unobserved carrier. Note properties of volatiles in protoplanetary disks. Their com- that the plot has been split into two different y-axis scales position is dominated by the classical carbon-nitrogen- to simultaneously show major as well as minor carriers. oxygen elemental trifecta as shown in Figure 6. In addition there are likely trace species including other nucleosyn- (only observed in the form of HCN) is still largely unac- thetic elements (e.g., sulfur and fluorine), although the only counted for by about 2 orders of magnitude. Carr and firm detections to date are of sulfur-bearing species, and Najita (2011) used a marginally different analysis method, those appear to indicate that the bulk of the sulfur is se- but found slightly higher HCN and C2H2 abundances by questered in a refractory carrier (Dutrey et al. 2011). factors of a few in similar, albeit not identical, disks. This, Comparing the observed column densities of CNO- while underlining that molecular abundances retrieved from bearing volatiles in the inner disks, Salyk et al. (2011) find low-resolution spectra are still uncertain, does not change that, while the data are consistent with most of the oxy- the basic demographic picture. The (slight) carbon deficit gen being carried by water and CO, carbon seems to be may be explained by some sequestration in a carbonaceous slightly under-abundant by a factor of 2–3, while nitrogen

11 refractory component, similar to the situation in the ISM. freeze-out. This is, however, not consistent with the carbon deficit in the inner solar system, as discussed in §2.2. The nitrogen 3.4. Chemical dependencies on stellar type deficit is potentially interesting as NH3 can be ruled out as a significant carrier in the inner disk by stringent upper There are large differences in the observational signa- limites on the presence of emission bands at 10 µm (Salyk tures of volatiles as a function of the mass and evolution- et al. 2011) and 3 µm (Mandell et al. 2012). This has led to ary stage of the central star. It is generally difficult to de- speculation that most of the nitrogen in the terrestrial region tect molecular emission from disks around Herbig Ae stars is stored in N2, and is largely unavailable for planetesimal (see Figure 7), and the implications of this strong observ- formation. Beyond the terrestrial region toward the outer able are currently being debated. The effect is strongest in disk, we would still expect to find nitrogen in the form of the inner disk, where very few disks around stars hotter than NH3 to explain the cometary and giant planet abundances. ∼ 7000 K show any molecular emission at all (Pontoppidan The inner disk elemental abundances are summarized in et al. 2010a), with the exception of CO (Blake and Boogert Table 1, and illustrated in Figure 6. 2004) and OH (Mandell et al. 2008; Fedele et al. 2011). The apparent increases of the OH/H2O ratio in Herbig stars have led to speculation that the strong dependence of molecular a Table 1: Distribution of C, N and O in inner disks. line detections on spectral type is due to the photodestruc- Oxygen Carbon Nitrogen tion of water in the photolayers of Herbig disks, rather than %%% an actual deficit of bulk midplane water. However, it is pos- H2O 38 – – sible that an enhanced abundance of the photo-products of CO 30 57 – water, and radicals in general, in combination with efficient b HCN – 0.1-0.6 0.4-2.5 vertical mixing, can lead to large changes in bulk chemistry NH3 –– < 1.1 dependent on spectral type. It should be noted, however, C2H2 – 0.03-0.25 – that it is still debated how much of a depletion the non- CO2 (0.04) (0.05) – detections actually correspond to, as the dynamic range of OH (0.04) – – the infrared Spitzer spectra is relatively low. A detection of Silicates 32 – – water in the disk around HD 163296 suggests that the de- Total 100c 57-58 0.4-2.5 pletion in the inner disk in some cases may be as little as 1–2 orders of magnitude, although in this case, the water a Values are relevant for disks around stars of roughly 0.5-1.5 M . emission originates from somewhat larger radii (15-20 AU) bRanges indicate differences in median derived values from than traced by the multitude of mid-infrared molecular lines Salyk et al. (2011) and Carr and Najita (2011). (∼1 AU, Fedele et al. 2012). cIt is assumed that all the oxygen is accounted for, and the other Toward the lower end of the stellar mass spectrum, Pas- fractions are scaled using the solar elemental abundances. cucci et al. (2009) find that disks around young stars with spectral types M5–M9 have significantly less HCN relative As a comparison to the elemental budgets determined to C H , possibly requiring a greater fraction of the elemen- from molecular lines, a number of optical integral-field 2 2 tal nitrogen being sequestered as highly volatile N . Further spectroscopic measurements of the abundances of many el- 2 emphasizing the dependence of stellar type on inner disk ements in the photoevaporative flows from protoplanetary chemistry, Pascucci et al. (2013) recently found strongly disks in Orion are available (Tsamis et al. 2011; Mesa- enhanced carbon chemistry in brown dwarf disks, as mea- Delgado et al. 2012; Tsamis et al. 2013). The Orion disks sured by HCN and C H , relative to oxygen chemistry, as are being eroded by the harsh interstellar radiation field 2 2 measured by H O. Observationally, solar-mass young stars from young massive stars, providing a unique opportunity 2 therefore seem to be the most water-rich, and research in to measure their bulk composition. They show almost no the near future is likely to explore how this may affect the depletion. In the cases of carbon, oxygen, the abundances structure and composition of planetary systems around dif- are within a factor 1.5 relative to solar, while the nitrogen ferent stars. abundance is a little more than a factor 2 lower than solar. This suggests that most of the unseen molecular carriers are 3.5. Origin of disk volatiles: Inheritance and reset being entrained in the photoevaporative flows. In the outer disk, the CNO budget is observationally still How much of the proto-stellar volatile material survives incomplete due to strong depletions from the gas-phase by incorporation into the disk? That is, are initial disk con- freeze-out. In general most molecules are depleted relative ditions predominantly protostellar (Visser et al. 2009) and to cloud abundances (Dutrey et al. 1997) and CO is under- to what degree do the initial chemical conditions survive abundant by about an order of magnitude (van Zadelhoff the physical evolution of protoplanetary disks? Which re- et al. 2001; Qi et al. 2004). Water is depleted by up to 6–7 gions of the disk are inherited giving rise to planetesimals of orders of magnitude (Bergin et al. 2010; Hogerheijde et al. circum- or interstellar origin, and which regions are chem- 2011) - possibly even more than can be explained by pure ically reset, leading to disk-like planetesimals? Chemical

12 Fig. 7.— The relation between Spitzer detection rates of the infrared molecular forest and stellar spectral type (from Pontoppidan et al. 2010a).

Fig. 8.— Comparison between the observed abundances models indicate that essentially all disk regions are chem- of gas-phase inner disk volatiles derived from Spitzer-IRS ically active on time scales shorter than the lifetime of the spectra (Salyk et al. 2011) relative to those in ices in proto- disk. The dominant chemical pathways and the timescales stellar clouds (Oberg¨ et al. 2011b; Lahuis and van Dishoeck for equilibriation, however, are expected to vary substan- 2000). Disk abundances are appropriate for the inner disk, tially from region to region, depending on the physical as the Spitzer-IRS emission lines originate primarily in the properties of the local environment (Semenov and Wiebe few AU region. 2011). Partial answers to these questions may be found in the observed volatile demographics and their comparison to protostellar chemistry. Different scenarios should result in pend on the exposure of disk material to conditions that fa- significant differences in volatile composition and isotope vor chemical reactions, i.e. UV and X-ray radiation, high fractionation. Searching for such differences or similarities densities and heat. Disk models that begin with mostly is the subject of broad ongoing observational investigations. atomic structures implicitly assume that the disk chemistry Prestellar and protostellar volatile compositions have is reset. This assumption is plausible for regions inside the been systematically investigated through gas and ice obser- snow line because of the short chemical timescales (104 vations (van Dishoeck 2004; Oberg¨ et al. 2011b; Caselli years or less) for regions dominated by gas-phase chemistry and Ceccarelli 2012). There is some familial resemblance (Woitke et al. 2009). This is consistent with the meteoritic between the protostellar and comet compositions in our So- evidence that has always favored a complete chemical re- lar System, which has been taken as evidence of a direct in- set; most of the collected meteorites originated from feed- heritance of the comet-forming volatile reservoir (Mumma ing zones within the solar snow line (see also Section 2 for a and Charnley 2011). The possibility of such a connec- discussion of the evidence for a reset from a cosmochemical tion is supported by recent two-dimensional hydrodynam- perspective). ical simulations of the chemical evolution on trajectories The addition of accretion flows and advection, which from protostellar envelopes up to incorporation into the transport volatiles from the outer to the inner disk, as well disks (Aikawa et al. 1999; Brinch et al. 2008; van Weeren as vertical and radial mixing driven by local turbulence, re- et al. 2009; Visser et al. 2009, 2011; Hincelin et al. 2013). sult in mixing between a chemically active surface and a While different trajectories are exposed to different levels relatively dormant midplane. Semenov and Wiebe (2011) of heat and radiation, most ices that are deposited into the explored the different turbulence and chemistry timescales comet-forming region survive incorporation, and only some in disks, deriving characteric timescales to determine which of the most volatile ices sublimate during disk formation chemical processes can be efficiently erased by mixing (e.g. N2 and CO). (with necessary assumptions about disk parameters, includ- Some ice evolution during disk formation is likely, how- ing ionization rates, structure, etc.). They found that vertical ever, since the grains spend a significant time at 20-40 K – mixing is faster than the disk lifetime (< 106 years) interior temperatures known to be favorable for complex molecule of 200 AU in a typical disk. Ion-molecule reaction rates formation (Garrod and Herbst 2006; Garrod et al. 2008; depend on the abundances of ions involved in the reaction, Oberg¨ et al. 2009). Once the disk is formed, the initial but can be fast; as a representative example, they estimate volatile abundances may evolve through gas and grain sur- rates leading to the production of HCO+ to be < 104 years face reactions. The importance of this in situ processing throughout the disk. In such a case, the gas-phase chem- compared to the chemistry preceding the disk stage will de- istry will not retain any memory of the original composi-

13 tion, and inside the snow line, it is very difficult to retain tents of their disks. Not surprisingly, the molecular emis- any primordial signature of volatiles over the disk life time. sion lines from eruptive variables show remarkable variabil- Photochemistry is fast outside of the midplane. Outside the ity throughout outbursting events. For example, observa- snowline, the midplane is dominated by fast freeze-out and tions of ro-vibrational CO in the EXor V1647 Ori revealed a slower desorption by cosmic rays that still has a timescale fundamental and overtone CO emission at the beginning of shorter than the disk lifetime, implying that some gas-grain its outburst (Vacca et al. 2004; Rettig et al. 2005), cooler circulation will take place. Grain surface chemistry can in emission and a blue-shifted absorption after return to pre- some cases be very slow, however, and thus icy grains that outburst luminosity, and finally a ceasing of all emission do not get dredged up into the disk atmosphere or into the after one year (Brittain et al. 2007). Variable rovibrational inner disk may retain an inherited chemistry. Gas-phase lines of CO have also been observed towards EX Lupi, show emission from disks should mainly reflect disk in situ chem- evidence for at least two gas components at different tem- istry, especially from regions inside the main condensation peratures, with one tightly correlated with the system lumi- fronts, including e.g. isotopic fractionation patterns, while nosity, and displaying asymmetries due to a possible disk comet compositions and probes of disk ices from beyond a hotspot, and evidence for a warm molecular outflow (Goto fews 10’s of AU should exhibit protostellar-like abundances et al. 2011). Recently, a number of additional molecular and fractionation patterns. species have been observed in EX Lupi both before and In situ deuterium fractionation in disks provides the most during an outburst taking place during 2008, viz. H2O, OH, obvious explanation for the recent observations of different C2H2, HCN and CO2 (Banzatti et al. 2012). When the star disk distributions of two deuterated molecules, DCN and underwent the outburst, the organic lines went away while DCO+, with formation pathways that present different tem- the water and OH lines brightened, suggestive of the simul- perature dependencies (Oberg¨ et al. 2012). As discussed taneous evaporation of large amounts of water locked in ice in Section 2.4, cometary H2O deuterium fractionation may pre-burst, and active photo-chemistry in the disk surface. or may not be consistent with protostellar values, while ni- 3.7. Was the solar nebula chemically typical? trogen fractionation as well as NH3 spin temperatures both point to single cold origin (Mumma and Charnley 2011), Although destined to grow dramatically in the future, which would require that part of the disk (most likely the comparisons of protoplanetary disk chemistry with aspects disk midplane) retain a history of the initial (cloud) condi- of the cosmochemical evidence from the solar nebula are tions. emerging. It indeed appears that critical structures in the So how do the chemical abundances of solar-type in- solar nebula with respect to volatiles are reproduced in pro- ner disks compare to those of protostellar clouds and en- toplanetary disks. The astrophysical perspective shows a velopes? Figure 8 shows a comparison of the abundances clear distribution of a volatile-rich inner nebula and volatile- of HCN, CO2,H2O and NH3 (relative to CO) derived from poor outer nebula, with different snowlines between water Spitzer-IRS observations of disks (Salyk et al. 2011) with and more volatile CO. those derived from protostellar clouds (Oberg¨ et al. 2011b). For instance, do the D/H fractionation patterns in pro- While the abundance of H2O is similar for both clouds and toplanetary disks match those of the solar system? While 18 disks, inner disk abundances of other observed species are still elusive in disks, HDO and H2 O have been detected lower than in clouds. In summary, evidence for signifi- in the warm envelopes in the immediate surroundings of cant changes in inner disk chemistry during disk formation young solar mass stars. The measurements have a wider and evolution is emerging, emphasizing that protoplanetary degree of diversity than observed in solar system materi- disks have inner regions strongly affected by a reset-type als, with ratios of warm HDO to H2O ranging from < chemistry. To what degree this is consistent with the cos- 6 × 10−4 (Jørgensen and van Dishoeck 2010) to as high mochemical record will be an active research subject in the as 6 − 8 × 10−2 (Coutens et al. 2012; Taquet et al. 2013; near future. corresponding to D/H ratios a factor of two lower). Part of the explanation for this heterogeneity could be that the 3.6. Volatiles during transient accretion events derivation of the D/H ratio depends sensitively on whether the observations resolve the innermost (presumably disk- Chemical models of disk evolution often assume a rel- like) regions (Persson et al. 2013a), as well as on the details atively quiescent disk, experiencing processes no more dy- of the retrieval method, with recent models deriving lower namic than viscous accretion. However, evidence is mount- values of D/H (Visser et al. 2013; Persson et al. 2013b). ing that accretion of material from the natal cloud to the Nevertheless, the lower values of D/H for many sources are disk must occur in short, intense bursts that may drive the similar to those of cometary ices. We direct the reader to chemical reset over a relatively wide range of disk radii (e.g. the chapter by van Dishoeck et al. for more details about Dunham et al. 2010; see also the chapter on young erup- the analysis and interpretation of water vapor observations tive variables elsewhere in this book). The outbursts of FU in young, embedded disks. Ori and EXor stars reveal violent and energetic accretion Protoplanetary disks also show tentative evidence in sup- episodes (e.g. Herbig 1966, 1977; Hartmann and Kenyon port of proposed solutions to the carbon deficit problem de- 1996) that must have profound effects on the volatile con-

14 scribed in Section 2.2. [C I] has not yet been detected in – the consensus is that this is not possible because suffi- disks, with strong upper limits suggesting a general deple- cient amounts of mass cannot be maintained within such tion of gas-phase carbon relative to the observed dust mass, small radii. Rather, the planets migrate radially through even after accounting for CO freeze-out (Panic´ et al. 2010; some combination of gas interactions during the protoplan- Chapillon et al. 2010; Casassus et al. 2013). While a possi- etary disk phase (Lin et al. 1996), dynamical interactions ble explanation for this depletion is that the overall gas/dust with a planetesimal cloud (Hahn and Malhotra 1999), and ratio of the disk is depleted, Bruderer et al. (2012) show three-body interactions with a stellar companion (Fabrycky that, for at least one disk – HD 100546 — a low gas/dust and Tremaine 2007). While a discussion of the dynamical ratio is inconsistent with observed CO line fluxes, and in- evolution of planetary systems is beyond the scope of this stead suggest that the disk gas is depleted in carbon. A nat- chapter, the important point, when relating exoplanets to ural way to preferentially deplete carbon in the disk gas is the volatile system in protoplanetary disks, is that observed to convert it into a less volatile form. This idea is also sup- planetary orbits cannot be assumed to reflect their birth radii ported by the first attempt to directly measure the CO/H2 and feeding zones. The chemical composition of their plan- ratio in a disk; using observations of HD and CO in the TW ets may have been set by disk conditions that were never Hya disk, Favre et al. (2013) derive a CO/H2 ratio a few present at their ultimate mean orbital radius. In the solar to 100 times lower than the canonical value, suggesting the system, it was difficult to deliver the Earth’s water, while conversion of much of the carbon in CO into more refrac- exo-planetary systems may have common water worlds that tory compounds, such as organics. formed beyond the snow line, but heated up either due to or- Protoplanetary disk observations have also been used bital or snow line migration (Tarter et al. 2007). to search for evidence for isotope-selective photodissoca- Planetary migration may be seen as a complication when tion of CO, predicted to be the cause of the solar system’s interpreting observations, but perhaps it is also an opportu- mass-independent oxygen anomalies (see Section 2.4). Us- nity. We review two different aspects relating disks to exo- ing high-resolution spectrosopic observations of protoplan- planets. One is the question of whether exoplanet architec- etary disks, Smith et al. (2009) directly measured the abun- ture (bulk mass and orbital structure) reflects the properties dances of C16O, C17O and C18O. They found evidence of the volatile distribution in their parent disks, even though for mass-independent fractionation, consistent with a sce- some planetary migration takes place. Another is whether nario in which the fractionation is caused by isotope selec- the chemical structure of exoplanet atmospheres constrain tive photodissocation in the disk surface due to CO self- their birth radii, and therefore provides an independent mea- shielding (Lyons and Young 2005; Visser et al. 2009). sure of the migration process by allowing a comparison be- tween the initial and final orbital radii. 4. Disk volatiles and exoplanets In Section 2, we discussed how the presence of a water 4.1. The importance of solid volatiles in planet forma- snow line shaped the composition of the different planets tion and other bodies in the solar system, depending on their for- Do condensible volatiles play a central role in the basic mation location and, perhaps, dynamical history. We also ability of protoplanetary disks to form planets of all types, demonstrated that molecular species other than water must or are they merely passengers as the rocky material and have had similar strong radial gas-to-solid gradients, as pre- H2/He gas drive planet formation? The answer to this ques- served in elemental and isotopic abundances. Now that we tion generally seems to be that condensible volatiles do play have an increasingly detailed, if not yet fully comprehen- a central role. As discussed in section 2.6, the total mass sive, picture of protoplanetary disk chemistry and thermo- of condensible volatiles available for planetesimal forma- dynamics, supported by astronomical data, it is natural to tion is at least twice that of rock, and maybe as much as ask how volatiles impact the architecture and composition 3–4 times as much, at radii beyond the main condensation of exoplanets. Planetary properties that may be influenced fronts. The large increase in solid mass offered by condens- by the natal volatile composition include system character- ing volatiles affects the growth of planets on all scales – in istics, such as the distribution of masses and orbital radii, grain-grain interactions, in the formation of planetesimals their bulk chemistry (the fractions of rock, ice and gas) and and in the formation of gas giants. their atmospheric and surface chemistry. At the microscopic level, icy grains may stick more effi- While the combination of inner, dry terrestrial planets ciently at higher collision velocities, than bare rocky grains. and outer icy giants and planetesimals in the solar system Simulations suggest that the break-up velocity for icy grains strongly suggest that the water snow line is irreversibly im- may be 10 times higher than that of silicate grains (Wada printed on the large-scale architecture of a planetary system, et al. 2009). Other mechanisms that aid in the growth of icy the discovery of hot-Jupiter planets on short-period orbits grains include the potential to melt and rapidly re-freeze universal showed that there is, in fact, no such relation- upon high velocity collisions, leading to sticking even in Mayor and Queloz Marcy and Butler ship ( 1995; 1996). energetic collisions (Wettlaufer 2010). The solution does not appear to be to allow giant plan- An increase in solid mass may also catalyze the for- ets to form close to the parent star (well within 1 AU) mation of planetesimals – a critical step without which no

15 planet formation could take place in any but the most mas- Condensation fronts evolve strongly with time as the sive disks. At sufficiently high solid surface densities, the disk temperature changes in response to evolving stellar in- so-called streaming instability can lead to the rapid forma- solation and internal disk heating (e.g. Lissauer 1987), and tion of planetesimals by gravitational collapse. This is due with stellar luminosity (Kennedy and Kenyon 2008). At to a combination of pressure gradient flows and drag forces the earliest stages, high accretion rates can push the snow exerted by solid particles back onto the gas (e.g., Youdin line out to 5 AU or more around a young solar mass star. and Shu 2002; Johansen et al. 2007). At intermediate stages (1-2 Myr) during which giant plan- In the context of giant planet formation by core accre- ets may accumulate most of their mass, disk accretion rates −9 −1 tion, Pollack et al. (1996) demonstrated that a factor of 2 in drop toward 10 M yr , decreasing viscous heating to solid surface density can lead to a reduction in giant planet the point where the snow line moves within 1 AU. At the formation times by a factor 30. The time scale is determined latest stages of evolution, the disk clears, and the snow line by the time it takes for a planet to reach the crossover mass radius is again pushed outwards as the disk becomes opti- – the mass at which runaway gas accretion occurs. Since cally thin in the radial direction exposing the disk midplane the lifetime of the gas in protoplanetary disks is compara- to direct stellar irradiation (Garaud and Lin 2007; Zhang ble to the time scale for effective core accretion, factors of et al. 2013). We show the predicted evolution of midplane a few in solid mass can make a large difference in whether snow line distances for a range of stellar masses in Figure a system forms giant planets or not. Planet formation mod- 9 (based on the models of Kennedy and Kenyon 2008). Im- els that include more complexity, such as orbital migration portantly, it is seen that, while the end-stage snow line is (Ida and Lin 2004a), and evolving disks across a wide pa- located well outside the habitable zone of a solar mass star rameter space (Mordasini et al. 2009), confirm this basic (where liquid water may exist on a planetary surface; Kast- conclusion and emphasize the need for solid-state volatiles ing et al. 1993), it was actually inside the habitable zone around solar-metallicity stars to form gas giants. during the critical period in the planet formation process The dependence of planet formation on the availabil- that sees the formation of giant planets and the planetesi- ity of solid mass is famously reflected in the positive re- mal system. Consequently, there is a possibility that habit- lation between stellar metallicity and the frequency of gas able water worlds are more common among exo-planetary giants with masses & MJup (Santos et al. 2004; Fischer systems than the solar system would initially suggest. and Valenti 2005), coupled with a lack of such a relation We can also consider the dependence of stellar mass on with Neptune-mass planets (Sousa et al. 2008). The inter- this basic form of the snow line evolutionary curve. While pretation of these observations is that low metallicity disks stars with M∗ & 1 M have disks that during the opti- cannot form giant planets within the disk life time because cally thick phase experience a snow line radius significantly the solid cores never grow sufficiently massive to accrete closer to the star than the optically thin end stage configura- large gas envelopes, while the Neptune-mass planets can tion, disks around lower-mass stars follow a different path. form over a wider range of stellar metallicity (Ida and Lin Around these stars, the optically thin end-stage snow line 2004b). is the closest it gets. This is due to the relative dominance of accretion heating versus direct stellar irradiation around 4.2. Condensation fronts: A generalized concept of the low luminosity low-mass stars, suggesting that plane- snow lines tary migration is more important for forming water worlds in low-mass systems. When searching for links between protoplanetary disk chemistry and exoplanets, it becomes necessary to gener- 4.3. Transport of volatiles alize the concept of the snow line. Any molecular, or even atomic, species with condensation temperatures between 10 Protoplanetary disks are evolving systems that transport and 2000 K will have a curve demarcating the border be- a variety of quantities, angular momentum, mass and energy yond which the partial vapor pressure drops dramatically (see Armitage 2011, for a general review). The transport (the species freezes out). Further, the border is rarely a and mixing of solids in protoplanetary disks is a large, gen- straight line, but a surface that roughly follows an isotherm, eral area of study, but one that has important consequences with some departures due to the pressure dependency on the for disk volatiles. Like refractory material, condensed- condensation temperature. In a disk, a condensation front phase volatiles can move, in bulk, relative to each other and is typically strongly curved and becomes nearly horizon- to the primary H/He gas reservoir. However, as opposed tal for disk radii dominated by external irradiation heating to the refractories, a property of volatiles that make them a (Meijerink et al. 2009), or as a pressure gradient effect in re- special case is their ability to co-exist in solid and gaseous gions that are mostly isothermal (Ros and Johansen 2013). forms and to go through rapid phase changes across rela- Use of the term condensation front to describe this surface tively sharp condensation fronts. Two different transport for a given molecular or atomic species allows the snow line mechanisms may act in conjunction to concentrate volatiles, nomenclature to be retained as a specific reference to water and they are therefore often discussed together: The move- freeze-out at the disk mid plane, a distinction we will utilize ment of gasous volatiles by diffusion across the steep partial in this text. pressure gradients of a condensation front and the migration

16 will diffuse out across the condensation front again where it re-freezes (mostly due to turbulent mixing, although other types of mixing may also operate). In the absence of in- wards advection, this can lead to the complete depletion of inner disk volatiles. The process of diffusion against a partial pressure gradient is known as the cold finger effect (Stevenson and Lunine 1988). Together, these effects – in- wards solid advection and outwards gas diffusion – may, for certain disk parameters, increase the solid volatile sur- face density just beyond the condensation fronts by an or- der of magnitude (Cuzzi and Zahnle 2004; Ciesla and Cuzzi 2006). An important consequence of the transport of volatile species toward condensation fronts located at different disk radii may be to impose a strong radial dependence on the el- emental C/O ratio, as suggested by models of meteorite for- mation (Hutson and Ruzicka 2000; Pasek et al. 2005). In di- rect analog to other astrochemical environments where the Fig. 9.— The evolution of the midplane snow line as a elemental C/O ratio dictates a water- or carbon-dominated function of stellar mass. The data are taken from Kennedy chemistry, disks are expected to be similarly sensitive. It and Kenyon (2008) and modified with an optically thin end may therefore be possible to measure the disk C/O ratios at stage due to disk dissipation. Aside from the final optically various radii by measuring the relative abundances of water thin stage, the shapes of the curves are determined by the and carbon-bearing volatile species. A range of HCN/H2O stellar luminosity and accretion history of the disk. Note ratios measures using Spitzer spectra (Salyk et al. 2011; that the 3 M star undergoes large changes in luminosity Carr and Najita 2011), and an observed correlation of as it approaches the main sequence, while the lower-mass HCN/H2O line strengths with disk mass is at least consis- stars never reach the main sequence during the disk life- tent with the possibility that planetesimal growth could se- time. The dashed line represents the predicted snowline for quester O at or beyond the snow line and increase the inner an optically thin disk around a sun-like star, originally de- disk C/O ratio (Najita et al. 2013). rived by Hayashi (1981). Vertical transport of volatiles, while involving less bulk mass than radial transport, may act to significantly alter the observables of disk volatiles. In the range of radii in the of boulder-sized icy bodies against a shallow bulk (H/He) disk where the condensation fronts are nearly horizontal gas pressure gradient. (and very close in height) due to direct stellar irradiation Radial pressure gradients in protoplanetary disks are of the disk, the cold finger effect may act in the vertical well-known catalysts of solid transport through advection. direction. That is, volatile vapor turbulently mixes across In the absence of other effects, they are sufficiently efficient the condensation front, below which it freezes out. Some to move large fractions of the entire solid mass in the disk. of the resulting solids will then settle toward the midplane, Such pressure gradients support the gas by a radial force leading to a net loss of volatiles from the disk surface. This component, allowing it to orbit at slightly sub- or supra- was used as a proposed, but qualitative, explanation for the Keplerian velocities, depending on the direction of the gra- observed water vapor distribution in the AS 205N disk, in dient. Gas that is not in a Keplerian orbit, in turn produces which Spitzer spectroscopy suggested a snow line near 1 a dissipative force on solid particles, causing them to lose AU, well within the ∼ 150 K region normally demarcating or gain angular momentum and migrate radially (Weiden- the snow line (Meijerink et al. 2009). The effect was sub- schilling 1977; Stepinski and Valageas 1997). The rate of sequently supported by the numerical models of Ros and advection is a sensitive function of particle size, with the Johansen (2013). highest rates seen for sizes of 1-100 cm. For viscously Beyond the condensation front, grain settling may also evolving disks, the pressure gradient is monotonically de- deplete the surface of all solid volatiles. This was sug- creasing with radius, leading to rapid inwards advection. gested to explain the extremely low abundances of water This mechanism has the power to dramatically change the vapor observed with HIFI onboard the Herschel Space Ob- surface density of, in particular, volatiles, as icy bodies are servatory in the outer TW Hya disk ( 50 AU) (Hoger- transported from the outer disk toward, and across the snow & heijde et al. 2011). While it may seem counter-intuitive line (Ciesla and Cuzzi 2006). to constrain the abundance of water ice using water vapor, As the ices cross over their respective condensation this works in the disk surface because ultra-violet and X- fronts and subsequently sublimate, the partial pressure of ray radiation will desorb water ice into the gas-phase at a each volatile species sharply increases. Such steep par- constant rate at a given incident intensity (Ceccarelli et al. tial pressure gradients are not stable, and the volatile gas 2005; Dominik et al. 2005). In the case of TW Hya, the

17 water vapor abundance was found to be 1-2 orders of mag- tational CO lines have proven to be more challenging than nitude lower than expected from photo-desorption, leading initially expected (Qi et al. 2008). The CO condensation to a suggestion that the water ice was depleted by a similar front tends to get washed out in line images due to the high factor through a process of preferential settling. Similarly opacity of low-J CO lines and filling-in by warm surface low amounts of water vapor are consistent with upper limits gas. Effectively, the disk atmosphere hides the depletion in measured by Bergin et al. (2010) around DM Tau. the disk midplane. The solution to this problem has been to use chemical tracers. That is, molecules that only exist 4.4. Observations of condensation fronts in abundance when the CO is absent from the gas phase and/or abundant in the solid phase, thus avoiding interfer- Based on the development of high fidelity mid-infrared ence by emission from CO-rich regions. Four molecules + spectroscopy targeting warm volatiles, including water, as have been suggested: N2H , which is efficiently destroyed well as some of the first ALMA line imaging results, it has by gas-phase CO (Qi et al. 2013a), H2CO and CH3OH, now become feasible to measure, and even image, the lo- which form from CO ice hydrogenation, and DCO+, which cation of condensation fronts. The observations address is enhanced as CO depletion allows the formation of one + specific technical challenges in measuring a condensation of its parent molecules, H2D (Mathews et al. 2013). Ex- front. amples of CO condensation fronts measured using DCO+ + One challenge is related to the small angular size of the (HD 163296) and N2H (TW Hya) are shown in Figure 10. + gas-rich region (10s of milli-arcseconds for water and ∼0.5 N2H ,H2CO and CH3OH should all be present exclu- arcseconds for CO in nearby star forming regions); to mea- sively outside of the CO snow line, with an inner edge that sure the snow line with a direct line image, the spatial res- traces its position. The case of DCO+ is complicated be- olution of the observation must be significantly better than cause CO is also a parent molecule, and its abundance is + these sizes. To characterize the shape and sharpness of the a result of a competition between CO and H2D , lead- condensation front, the resolution must be even higher. As ing to efficient formation in a narrow temperature range an alternative to direct imaging, the many transitions of wa- just around the CO condensation front. Recent images of ter covering virtually all excitation temperatures provides DCO+ in the disk around HD 163296 show a ring-like an opportunity to spectrally map its radial distribution in structure, with a radius consistent with that expected for CO a disk, using temperature as a proxy for radius. The first condensation, suggestive that at least for this disk, DCO+ qualitative attempt was made by Meijerink et al. (2009) for is a good CO snow line tracer (Mathews et al. 2013, Fig- the disk around AS 205N, and the method was refined and ure 10). However, TW Hya does not show the same ring- quantified for the transitional disk TW Hya by Zhang et al. like structure in DCO+ (Qi et al. 2008) while a more recent + (2013). ALMA image of N2H in this disk confirms a clear ring- In TW Hya, the inner few AU are depleted of small dust like structure, with a location consistent with the expected grains (Calvet et al. 2002), and the snow line is located fur- radius of the CO snow line (Qi et al. 2013b, Figure 10). ther from the star than would be expected for a classical optically thick disk. Instead, the decreased radial optical depth of grains allows radiation to penetrate to larger radii in the disk, and moves the snowline to just beyond the tran- sition radius where the disk becomes optically thick. This observation is evidence that the snow line location moves outwards near the end of the gas-rich, optically thick phase of protoplanetary disk evolution. One challenge to the temperature mapping method is that the disk midplane is shielded, either by highly opti- cally thick dust or by interfering line emission from a super- heated surface layer. For condensation fronts at small radii Fig. 10.— The CO snowline as observed with ALMA. The and relatively high temperatures, such as for water, the rel- CO snowline is observed using chemical tracers of the ab- evant mid- to far-infrared lines are therefore formed in the sence of CO from the gas-phase (left, DCO+ Mathews et al. disk surface above one scale height (Woitke et al. 2009), 2013 and right, N H+ Qi et al. 2013b). The dashed curves as opposed to in the midplane. In practice, this may skew 2 delineate the 17 K isotherms, where CO freezes out. measured condensation fronts to larger radii. While water has the most famous condensation front, others, characterized by lower temperatures, are large 4.5. The influence of disk volatiles on planetary com- enough to image directly with facilities like ALMA. Evi- position dence for CO freeze-out has been previously suggested due to a measured overall depletion of CO in disks relative to It is a basic expectation that the bulk elemental compo- dark cloud values (e.g. Qi et al. 2004). However, attempts to sition of a planet reflects that of protoplanetary disk mate- image the CO condensation front directly in disks using ro- rial that formed it. Further, as planets never fully mix and

18 equilibrate, interiors and atmospheres may also reflect dif- matics of protoplanetary disks would be far less advanced; ferences in the composition of material accreted during dif- we rely on molecular line tracers to measure velocities, tem- ferent stages of formation. This dependence is supported peratures and mass of the gas. The problem with using by the wide diversity of planets (and likely, moons) in ex- volatiles as physical tracers arises because the disks are far oplanetary systems; clearly, not all planets are the same, from well-mixed, chemically simple systems. Thus, kine- and to what degree is this due to disk volatiles and the matic, thermal and abundance measurements derived from processes that sculpt their distribution in a protoplanetary molecular lines depend on a model of the volatile system; disk? We are just beginning to be able to study the com- are most of the tracer molecules frozen out? Are we trac- positions of exoplanets as well, and diversity appears to be ing a specific layer where chemical formation happens to be the rule, rather than the exception. Thanks to the success more effective? Have the total gas+solid volatile reservoir of the Kepler mission (Borucki et al. 2010), other exoplanet been depleted due to transport? transit studies (e.g., Barge et al. 2008) and radial velocity In this section, we discuss the links between the physi- surveys (Mayor et al. 2003; Cumming et al. 2008) com- chemical behavior of disk volatiles and their use as phys- bined with transit observations, we can now determine the ical tracers, and in particular point out areas where our average density of some exoplanets, and study their bulk improved understanding of volatiles improve the physical chemistry (e.g., Sasselov 2003; Burrows et al. 2007). measurements. A more general discussion of measurements Further, the first observations of the atmospheric compo- of physical properties of protoplanetary disks can be found sitions of exoplanets are being made possible with studies elsewhere in this book. of primary and secondary planetary transits (Knutson et al. 2007), as well as direct imaging (Konopacky et al. 2013). 5.1. The impact of volatiles on disk mass measure- However, the relatively small number of planets that have ments been studied have atmospheric signatures that are not triv- ially reproduced by existing atmospheric models (e.g., Bar- The bulk gas masses and the gas-to-dust ratios of pro- man et al. 2011; Skemer et al. 2012). toplanetary disks are some of the most fundamental obser- One important parameter, among others, in modeling ex- vational parameters. They are often used to place the disks oplanet atmospheres is the elemental C/O ratio. The possi- into a global evolutionary scheme, or to compare their prop- ble detection of a high C/O ratio in the exoplanet WASP- erties to those inferred for the earliest stages of solar system 12b prompted a study suggesting a simple way in which formation. the disk architecture might affect the composition of plane- The bulk mass of disks has most commonly been mea- tary atmospheres (Madhusudhan et al. 2011). Oberg¨ et al. sured by observing the dust mass and converting to a gas (2011a) suggested that, if a giant planet atmosphere does mass. More recently, attempts have been made to measure not mix, or mixes incompletely with volatiles sequestered the gas mass directly, via various gas-phase tracers. How- in its core, it is possible to create an atmosphere with a high ever, no matter the method of bulk mass measurement, an C/O ratio near unity. This is possible in regions of the disk understanding of the volatile content of the disk is required, where oxygen preferentially condenses relative to carbon as the volatiles represent most of the disk mass. Therefore, (for instance outside the water condensation front, but in- many of the intrinsic properties of volatiles discussed else- side the CH4, CO2 and CO fronts). If planets form via core where in this chapter also create difficulties for attempts to accretion from solids followed by the accretion of a gaseous measure bulk disk masses. Nevertheless, there has been sig- envelope, their atmospheres will inherit the high C/O ratio, nificant progress on this front, which we describe here. driving an atmospheric carbon-dominated chemistry. This The use of the dust continuum emission is the oldest predicts different planetary atmospheric compositions as a method to estimate the total disk mass, and is still the function of initial formation radii, but independently of the most frequently used (Beckwith et al. 1990; Andrews and final planetary orbits. Williams 2005; Mann and Williams 2010). In converting from dust mass to total disk mass, the gas-to-dust ratio 5. The physical structure and kinematics of protoplan- and dust opacities are typically assumed. However, both of etary disks as traced by volatiles these assumed values are traditionally based on dust models that often do not include any of the CNOH volatiles. Are Our knowledge of the physical structure and kinematics these assumptions actually reasonable, given the knowledge of protoplanetary disks is fundamentally dependent on how that the disk dust mass is dominated by ices? For disk ma- well we understand the behavior of their volatiles. By their terial in the limiting (but well-mixed) case that everything, nature, volatiles can be found in the gas-phase throughout except for hydrogen and helium, is in the solid state, the the disk. Indeed, the total mass of protoplanetary disks is solar abundances of Grevesse et al. (2010) imply a gas-to- entirely dominated by a volatile molecule – H2. As we dust mass ratio of 72. The consequence is that disk masses have seen in the previous sections, much of the most abun- are probably slightly lower than if the gas-to-dust ratio is dant heavy elements are also often found in the form of gas- assumed to have the canonical value of 100, but only by a phase volatiles. This is fortunate because, in their absence, relatively small fraction. In terms of the dust opacity, the our current knowledge of the physical structure and kine- addition of ices has a complex behavior due to several com-

19 peting effects. Ice itself generally has lower mass opacity 5.2. The kinematics of disk volatiles than refractory dust, so the opacity of bulk disk material will decrease significantly with the addition of ice. Secondary The bulk mass of protoplanetary disks resides close to effects, such as a potentially increased sticking efficiency the midplane, and can be described by a Keplerian flow; may change the opacity of icy grains by catalyzing grain known departures from this, for instance due to radial growth. All together, the net effect seems to be to slightly pressure support (Weidenschilling 1977) are very slight (. decrease the mass opacity of ice by 20–30% (Ossenkopf 0.01VKepler), and still beyond our observational capabili- and Henning 1994; Semenov et al. 2003). Thus, volatiles ties. The surface layers of the disk, however, may be subject actually have a relatively minor effect on determination of to much greater dynamical effects in the form of enhanced bulk dust masses in disks. The one potential caveat is if ionization-dependent turbulence and winds. Some chemi- ices convert from simple species to complex CNOH organ- cal models now begin to consider the efficiency of process- ics, as suggested by solar system meteoritics. The organics ing volatiles as they are mixed into a turbulent surface layer and exposed to high temperatures and direct irradiation be- have much higher opacities than, say water, CO2 or NH3 ice, and may dominate the total dust opacity at high abun- fore being returned to the midplane (Semenov and Wiebe dances (Pollack et al. 1994). 2011; Furuya et al. 2013). As all chemical time scales in One approach to estimate disk masses with gas-phase the midplane are much longer than the photo-chemical time tracers has been to observe multiple rotational transitions of scale on the disk surface, such mixing can lead to significant multiple isotopologues of CO with millimeter interferome- changes to the bulk chemistry, if efficient. ters. The more optically thin isotopologues are sensitive Volatile emission lineshapes also hint at the tidal influ- to the disk mass, while the optically thick transitions (es- ence of planets. Models of forming gas giant planets em- pecially from higher-excitation states) are sensitive to the bedded in circumstellar disks indicate that their interactions disk temperature structure. Using this method, Panic´ et al. cause the inner edge of the cleared region exterior to their (2008) simultaneously measure the gas and dust masses in orbit to grow eccentric (Papaloizou et al. 2001; Kley and the disk around HD 169142 and find a gas/dust ratio in the Dirksen 2006; Lubow 1991). When the companion mass range of 25–100. grows to more than 0.3% of the mass of the central star, A new method for estimating disk mass was explored simulations show that azimuthally averaged eccentricities by Bergin et al. (2013), who used observations of HD can grow as large as e = 0.25 at the inner edge of the disk −2 J = 1 − 0 obtained with Herschel-PACS to measure the yet fall off as fast as r (Kley and Dirksen 2006). Gas disk mass around TW Hya. HD represents an attractive arising from material in a Keplerian orbit with a non-zero mass tracer for two main reasons: It directly probes the eccentricity reveals a distinct shape (Liskowsky et al. 2012) due to the uneven illumination of the inner edge of the disk bulk molecular component (H2) with a conversion factor and a slight Doppler shift relative to a circular orbit due to from HD to H2 expected to be near the interstellar value for nearly all regions of the disk, and, HD, unlike CO, is the non-circular velocities. Regaly´ et al. (2010) describe the not expected to freeze out in disk conditions. However, ro-vibrational CO line profiles that could arise from a disk as for all gas tracers, the result is still model-dependent harboring a supra-jovian mass companion. However, the through the assumed gas temperature. Bergin et al. (2013) CO emission originates from a large radial extent, diluting used observations of CO millimeter transitions to indepen- the signal from the asymmetric inner rim (Liskowsky et al. dently constrain the gas temperature distribution in the disk. 2012). A better tracer of the inner rim is ro-vibrational OH, They found a disk mass consistent with a gas/dust ratio near which may be excited preferentially in a ring at the inner 100, suggesting that even this evolved disk has a canoni- edge of the outer disk. Such asymmetric ro-vibrational OH cal gas/dust ratio and sufficient mass to form giant plan- emission has been observed toward V380 Ori (Fedele et al. ets. Measurements of the gas mass in TW Hya were also 2011) and HD 100546 (Liskowsky et al. 2012). It should attempted by Thi et al. (2010) and Gorti et al. (2011), be noted that other scenarios could produce asymmetries; both of whom used a large suite of emission lines, includ- for example, fluctuations of turbulence (e.g. Fromang and ing millimeter-wave CO lines, to constrain the disk prop- Nelson 2006) could give rise to non-axisymmetric den- erties. However, the model dependencies in this approach sity structures that result in asymmetric emission from OH, were exposed by the discrepant results produced by the two which originates in a region near the critical density for groups, which allowed for TW Hya to have either a very thermalizing its vibrational levels. Further temporal stud- sub-canonical or nearly canonical gas/dust ratio. Use of the ies of this effect are in progress to distinguish between the method presented by Thi et al. (2010) also gives gas-to-dust possible scenarios. ratios significantly below 100 for other disks (e.g., Tilling et al. 2012), demonstrating a strong need to reach a consen- 6. Future directions sus using different methods for disk mass measurements. Our understanding of protoplanetary disk evolution and planet formation is in a transitional stage. Where low resolution observations once led to models of disks as monolithic objects, characterized by a few scalar parame-

20 ters (mass, size, dust grain size, etc.), disks are now seen dramatically change the local mode of chemistry by as well-resolved composite entities with different, highly changing the C/O ratio and increase the local solid physically and chemically distinct regions. We are no surface density to aid in the formation of planetesi- longer averaging properties over entire disks, but are taking mals and planets. The near future is likely to bring the first steps toward deconstructing them by zooming in on observational searches for bulk transport of volatiles specific sub-regions to derive their causal interrelations. In that test dynamo-chemical models of protoplanetary this chapter, we have shown how volatiles play a key role in disks. sculpting a changing environment as we move from the in- ner terrestrial region of protoplanetary disks, across the gi- • Planet composition as a function of volatile con- ant planet region and toward the Kuiper belt and cometary tent: The composition of individual planets must regions. We expect that during the next 5-10 years, we will necessarily reflect that of the disk material used in see dramatic progress in our understanding of disk chem- their formation. Thus, there is an expectation that istry, the transport of bulk volatiles and their links to exo- the chemical demographics of disks will match the planetary systems: compositional demographics of exoplanetary sys- tems. The available data are still far too insufficient • Elemental budgets in disks: One of the most direct, to make such connections, but the rapid advances in and observationally feasible, comparisons of proto- both disk and planet observations are likely to soon planetary disks to the conditions in the solar nebula, make such comparisons common. is the distribution of elements in volatile and refrac- tory molecules. We expect that significant progress Acknowledgments. This work was supported by a in detecting all the bulk elemental volatile carriers NASA Origins of the Solar System Grant No. OSS 11- throughout protoplanetary disks will be made in the OSS11-0120, a NASA Planetary Geology and Geophysics next years. However, the success of this task will de- Program under grant NAG 5-10201, and by the National pend not only on new data, but also on improving the Science Foundation under grant AST-99-83530. The au- accuracy with which radiative transfer calculations thors are grateful to Ewine van Dishoeck and Thomas Hen- can retrieve the abundances. An important question is ning for comments that greatly improved this chapter. whether our inability to observe N will result in the 2 REFERENCES nitrogen budget remaining elusive for the foreseeable future? Aikawa Y. et al. (1999) Astrophys. J., 519, 705. Aikawa Y. et al. (2002) Astron. Astrophys., 386, 622. • Inheritance or reset? A central question is that of Aikawa Y. et al. (2012) Astron. Astrophys., 538, A57. inheritance. To what degree is the volatile composi- Albarede` F. et al. (2013) Icarus, 222, 44. tion transferred into disks from the dense interstellar Aleon´ J. (2010) Astrophys. J., 722, 1342. medium through the protostellar phase, and to what Alexander C. M. O. . et al. (2012) Science, 337, 721. degree are the volatiles formed in situ within chem- Allegre` C. et al. (2001) Earth and Planetary Science Letters, 185, ically active regions in the disk, resetting previous 49. Amelin Y. et al. (2002) Science, 297, 1678. chemical signatures and losing memory of the inter- Andrews S. M. and Williams J. P. (2005) Astrophys. J., 631, 1134. stellar phase. Likely, the answer will be that both Armitage P. J. (2011) Annu. Rev. Astron. Astrophys., 49, 195. types of evolution play a role with the latter being Banzatti A. et al. (2012) Astrophys. J., 745, 90. more important for the inner disk. Improvements in Barge P. et al. (2008) Astron. Astrophys., 482, L17. the accuracy with which chemical abundances can be Barman T. S. et al. (2011) Astrophys. J., 733, 65. retrieved as a function of location in the disk, with Beckwith S. V. W. et al. (1990) Astron. J., 99, 924. spatial fidelities as high as a few AU are required. Bergin E. A. et al. (2010) Astron. Astrophys., 521, L33. This will be provided by a synergistic combination Bergin E. A. et al. (2013) Nature, 493, 644. of submillimeter interferometric line imaging with Blake G. A. and Boogert A. C. A. (2004) Astrophys. J. Lett., 606, ALMA with mid- to far-infrared spectroscopy with L73. JWST and a range of ground-based optical/IR tele- Bockelee-Morvan´ D. et al. (2009) Astrophys. J., 696, 1075. Bockelee-Morvan´ D. et al. (2012) Astron. Astrophys., 544, L15. scopes. Importantly, no single observatory will be Boogert A. C. A. et al. (2008) Astrophys. J., 678, 985. sufficient in isolation because of the need to cover Borucki W. J. et al. (2010) Science, 327, 977. a wide range in gas temperatures and corresponding Briani G. et al. (2009) in: Lunar and Planetary Institute Science wavelength ranges. Conference Abstracts, vol. 40 of Lunar and Planetary Institute Science Conference Abstracts, p. 1642. • Transport of volatiles: Models have long predicted Brinch C. et al. (2008) Astron. Astrophys., 489, 617. that the local concentration of volatiles may vary by Brittain S. et al. (2007) Astrophys. J. Lett., 670, L29. orders of magnitude in different disk regions due to Brown J. M. et al. (2013) Astrophys. J., 770, 94. a combination of gas diffusion and solid particle ad- Brown M. E. et al. (2012) Astron. J., 143, 146. vection. Collectively, these transport processes may Brownlee D. et al. (2006) Science, 314, 1711. alter the observational signature of the disk surface, Bruderer S. et al. (2012) Astron. Astrophys., 541, A91.

21 Burnett D. S. et al. (2003) Space Sci. Rev., 105, 509. Grossman L. (1972) Geochim. Cosmochim. Acta, 36, 597. Burrows A. et al. (2007) Astrophys. J., 661, 502. Hahn J. M. and Malhotra R. (1999) Astron. J., 117, 3041. Bus S. J. and Binzel R. P. (2002) Icarus, 158, 1, 146 . Hama T. et al. (2011) Astrophys. J. Lett., 738, L15. Calvet N. et al. (2002) Astrophys. J., 568, 1008. Hartmann L. and Kenyon S. J. (1996) Annu. Rev. Astron. Astro- Campins H. et al. (2010) Nature, 464, 1320. phys., 34, 207. Carr J. S. and Najita J. R. (2008) Science, 319, 1504. Hartogh P. et al. (2011) Nature, 478, 218. Carr J. S. and Najita J. R. (2011) Astrophys. J., 733, 102. Hashizume K. et al. (2004) Astrophys. J., 600, 480. Carr J. S. et al. (2004) Astrophys. J., 603, 213. Hayashi C. (1981) Progress of Theoretical Physics Supplement, Casassus S. et al. (2013) Astron. Astrophys., 553, A64. 70, 35. Caselli P. and Ceccarelli C. (2012) Astron. Astrophys. Rev., 20, 56. Henning T. and Semenov D. (2013) Chemical Reviews, 113, 9016. Cassen P. (1996) Meteoritics and Planetary Science, 31, 793. Herbig G. H. (1966) Vistas in Astronomy, 8, 109. Ceccarelli C. et al. (2005) Astrophys. J. Lett., 631, L81. Herbig G. H. (1977) Astrophys. J., 217, 693. Ceccarelli C. et al. (2007) Protostars and Planets V, pp. 47–62. Hersant F. et al. (2001) Astrophys. J., 554, 391. Chapillon E. et al. (2010) Astron. Astrophys., 520, A61. Hincelin U. et al. (2013) Astrophys. J., 775, 44. Chapman C. R. (1996) Meteoritics and Planetary Science, 31, Hogerheijde M. R. et al. (2011) Science, 334, 338. 699. Honda M. et al. (2009) Astrophys. J. Lett., 690, L110. Chapman C. R. et al. (1975) Icarus, 25, 104. Horner J. et al. (2007) Earth Moon and Planets, 100, 43. Chiang E. I. and Goldreich P. (1997) Astrophys. J., 490, 368. Hutson M. and Ruzicka A. (2000) Meteoritics and Planetary Sci- Chiang E. I. et al. (2001) Astrophys. J., 547, 1077. ence, 35, 601. Ciesla F. J. and Cuzzi J. N. (2006) Icarus, 181, 178. Ida S. and Lin D. N. C. (2004a) Astrophys. J., 604, 388. Clayton R. N. (2002) Nature, 415, 860. Ida S. and Lin D. N. C. (2004b) Astrophys. J., 616, 567. Clayton R. N. et al. (1973) Science, 182, 485. Jacquet E. and Robert F. (2013) Icarus, 223, 722. Coutens A. et al. (2012) Astron. Astrophys., 539, A132. Jenniskens P. et al. (1993) Astron. Astrophys., 273, 583. Cumming A. et al. (2008) PASP, 120, 531. Jessberger E. K. et al. (1988) Nature, 332, 691. Cuzzi J. N. and Zahnle K. J. (2004) Astrophys. J., 614, 490. Johansen A. et al. (2007) Nature, 448, 1022. D’Alessio P. et al. (1997) Astrophys. J., 474, 397. Jones T. D. et al. (1990) Icarus, 88, 172. Dasgupta R. and Hirschmann M. M. (2010) Earth and Planetary Jørgensen J. K. and van Dishoeck E. F. (2010) Astrophys. J. Lett., Science Letters, 298, 1. 725, L172. Desch S. J. (2007) Astrophys. J., 671, 878. Jura M. (2006) Astrophys. J., 653, 613. Dodson-Robinson S. E. et al. (2009) Icarus, 200, 672. Jura M. (2008) Astron. J., 135, 1785. Dominik C. et al. (2005) Astrophys. J. Lett., 635, L85. Jura M. et al. (2012) Astrophys. J., 750, 69. Draine B. T. (2003) Annu. Rev. Astron. Astrophys., 41, 241. Kamp I. et al. (2013) Astron. Astrophys., 559, A24. Drouart A. et al. (1999) Icarus, 140, 129. Kasting J. F. et al. (1993) Icarus, 101, 108. Dunham M. M. et al. (2010) Astrophys. J., 710, 470. Kennedy G. M. and Kenyon S. J. (2008) Astrophys. J., 673, 502. Dutrey A. et al. (1997) Astron. Astrophys., 317, L55. Kerridge J. F. (1985) Geochimica et Cosmochimica Acta, 49, 8, Dutrey A. et al. (2011) Astron. Astrophys., 535, A104. 1707 . Ebel D. S. and Grossman L. (2000) Geochim. Cosmochim. Acta, Kitamura Y. and Shimizu M. (1983) Moon and Planets, 29, 199. 64, 339. Kley W. and Dirksen G. (2006) Astron. Astrophys., 447, 369. Epstein S. et al. (1987) Nature, 326, 477. Knutson H. A. et al. (2007) Nature, 447, 183. Fabrycky D. and Tremaine S. (2007) Astrophys. J., 669, 1298. Konopacky Q. M. et al. (2013) Science, 339, 1398. Farihi J. et al. (2013) Science, 342, 218. Kress M. E. et al. (2010) Advances in Space Research, 46, 44. Favre C. et al. (2013) Astrophys. J. Lett., 776, L38. Kuiper G. P. (1953) Proceedings of the National Academy of Sci- Fedele D. et al. (2011) Astrophys. J., 732, 106. ence, 39, 1153. Fedele D. et al. (2012) Astron. Astrophys., 544, L9. Lahuis F. and van Dishoeck E. F. (2000) Astron. Astrophys., 355, Fischer D. A. and Valenti J. (2005) Astrophys. J., 622, 1102. 699. Flasar F. M. et al. (2005) Science, 307, 1247. Lahuis F. et al. (2006) Astrophys. J. Lett., 636, L145. Frick U. et al. (1979) pp. 1961–1972. Lebofsky L. A. (1980) Astron. J., 85, 573. Fromang S. and Nelson R. P. (2006) Astron. Astrophys., 457, 343. Lebofsky L. A. et al. (1981) Icarus, 48, 453. Furuya K. et al. (2013) ArXiv e-prints. Lee J.-E. et al. (2010) Astrophys. J. Lett., 710, L21. Gail H.-P. (2002) Astron. Astrophys., 390, 253. Levison H. F. and Duncan M. J. (1997) Icarus, 127, 1, 13 . Gansicke¨ B. T. et al. (2012) Mon. Not. R. Astron. Soc., 424, 333. Levison H. F. et al. (2010) Science, 329, 187. Garaud P. and Lin D. N. C. (2007) Astrophys. J., 654, 606. Lin D. N. C. et al. (1996) Nature, 380, 606. Garrod R. T. and Herbst E. (2006) Astron. Astrophys., 457, 927. Liskowsky J. P. et al. (2012) Astrophys. J., 760, 153. Garrod R. T. et al. (2008) Astrophys. J., 682, 283. Lissauer J. J. (1987) Icarus, 69, 249. Geiss J. (1987) Astron. Astrophys., 187, 859. Lodders K. (2003) Astrophys. J., 591, 1220. Gerakines P. A. et al. (1996) Astron. Astrophys., 312, 289. Lord III H. C. (1965) Icarus, 4, 279. Gibb E. L. et al. (2004) Astrophys. J. Suppl., 151, 35. Lubow S. H. (1991) Astrophys. J., 381, 259. Gibb E. L. et al. (2007) Astrophys. J., 660, 1572. Luck J. M. et al. (2001) Meteoritics and Planetary Science Sup- Gomes R. et al. (2005) Nature, 435, 466. plement, 36, 118. Gorti U. et al. (2011) Astrophys. J., 735, 90. Lyons J. R. and Young E. D. (2005) Nature, 435, 317. Goto M. et al. (2011) Astrophys. J., 728, 5. Madhusudhan N. et al. (2011) Nature, 469, 64. Grevesse N. et al. (2010) Astrophys. Space Sci., 328, 179. Malfait K. et al. (1998) Astron. Astrophys., 332, L25.

22 Mandell A. M. et al. (2008) Astrophys. J. Lett., 681, L25. Qi C. et al. (2013a) Astrophys. J., 765, 34. Mandell A. M. et al. (2012) Astrophys. J., 747, 92. Qi C. et al. (2013b) Science, 341, 630. Mann R. K. and Williams J. P. (2010) Astrophys. J., 725, 430. Regaly´ Z. et al. (2010) Astron. Astrophys., 523, A69. Marcy G. W. and Butler R. P. (1996) Astrophys. J. Lett., 464, L147. Remusat L. et al. (2009) Astrophys. J., 698, 2087. Marrocchi Y. et al. (2005) Earth and Planetary Science Letters, Rettig T. W. et al. (2005) Astrophys. J., 626, 245. 236, 569. Rivkin A. S. and Emery J. P. (2010) Nature, 464, 1322. Marty B. (1995) Nature, 377, 326. Rivkin A. S. et al. (2002) Asteroids III, pp. 235–253. Marty B. (2012) Earth and Planetary Science Letters, 313, 56. Ros K. and Johansen A. (2013) Astron. Astrophys., 552, A137. Marty B. et al. (2008) Science, 319, 75. Salyk C. et al. (2008) Astrophys. J. Lett., 676, L49. Marty B. et al. (2010) Geochim. Cosmochim. Acta, 74, 340. Salyk C. et al. (2011) Astrophys. J., 731, 130. Mathews G. S. et al. (2013) ArXiv e-prints. Sandford S. A. et al. (2006) Science, 314, 1720. Mayor M. and Queloz D. (1995) Nature, 378, 355. Santos N. C. et al. (2004) Astron. Astrophys., 415, 1153. Mayor M. et al. (2003) The Messenger, 114, 20. Sasselov D. D. (2003) Astrophys. J., 596, 1327. McClure M. K. et al. (2012) Astrophys. J. Lett., 759, L10. Savage B. D. and Sembach K. R. (1996) Annu. Rev. Astron. Astro- McCord T. B. et al. (1971) Astrophys. J., 165, 413. phys., 34, 279. McKeegan K. D. et al. (2006) Science, 314, 1724. Scott E. R. D. (2007) Annual Review of Earth and Planetary Sci- McKeegan K. D. et al. (2011) Science, 332, 1528. ences, 35, 577. McNaughton N. J. et al. (1981) Nature, 294, 639. Semenov D. and Wiebe D. (2011) Astrophys. J. Suppl., 196, 25. Meijerink R. et al. (2009) Astrophys. J., 704, 1471. Semenov D. et al. (2003) Astron. Astrophys., 410, 611. Mesa-Delgado A. et al. (2012) Mon. Not. R. Astron. Soc., 426, Semenov D. et al. (2008) Astrophys. J. Lett., 673, L195. 614. Skemer A. J. et al. (2012) Astrophys. J., 753, 14. Min M. et al. (2005) Icarus, 179, 158. Sliter R. et al. (2011) J. Phys. Chem. A, 115, 9682. Mordasini C. et al. (2009) Astron. Astrophys., 501, 1139. Smith R. L. et al. (2009) Astrophys. J., 701, 163. Mousis O. et al. (2000) Icarus, 148, 513. Sousa S. G. et al. (2008) Astron. Astrophys., 487, 373. Mumma M. J. and Charnley S. B. (2011) Annu. Rev. Astron. As- Stepinski T. F. and Valageas P. (1997) Astron. Astrophys., 319, trophys., 49, 471. 1007. Mumma M. J. et al. (1988) in: Bulletin of the American Astronom- Stevenson D. J. (1985) Icarus, 62, 4. ical Society, vol. 20 of Bulletin of the American Astronomical Stevenson D. J. and Lunine J. I. (1988) Icarus, 75, 146. Society, p. 826. Taquet V. et al. (2013) ArXiv e-prints. Najita J. et al. (2003) Astrophys. J., 589, 931. Tarter J. C. et al. (2007) Astrobiology, 7, 30. Najita J. R. et al. (2013) Astrophys. J., 766, 134. Terzieva R. and Herbst E. (2000) Mon. Not. R. Astron. Soc., 317, Niemann H. B. et al. (1996) Science, 272, 846. 563. Notesco G. et al. (1999) Icarus, 142, 298. Thi W. F. et al. (2002) Astron. Astrophys., 394, L27. Oberg¨ K. I. et al. (2005) Astrophys. J. Lett., 621, L33. Thi W.-F. et al. (2004) Astron. Astrophys., 425, 955. Oberg¨ K. I. et al. (2009) Astron. Astrophys., 504, 891. Thi W.-F. et al. (2010) Astron. Astrophys., 518, L125. Oberg¨ K. I. et al. (2010) Astrophys. J., 720, 480. Thiemens M. H. and Heidenreich III J. E. (1983) Science, 219, Oberg¨ K. I. et al. (2011a) Astrophys. J. Lett., 743, L16. 1073. Oberg¨ K. I. et al. (2011b) Astrophys. J., 740, 109. Tilling I. et al. (2012) Astron. Astrophys., 538, A20. Oberg¨ K. I. et al. (2012) Astrophys. J., 749, 162. Tsamis Y. G. et al. (2011) Mon. Not. R. Astron. Soc., 412, 1367. Ossenkopf V. and Henning T. (1994) Astron. Astrophys., 291, 943. Tsamis Y. G. et al. (2013) Mon. Not. R. Astron. Soc., 430, 3406. Ott U. et al. (1981) Geochim. Cosmochim. Acta, 45, 1751. Vacca W. D. et al. (2004) Astrophys. J. Lett., 609, L29. Owen T. et al. (2001) Astrophys. J. Lett., 553, L77. van Dishoeck E. F. (2004) Annu. Rev. Astron. Astrophys., 42, 119. Panic´ O. et al. (2008) Astron. Astrophys., 491, 219. van Weeren R. J. et al. (2009) Astron. Astrophys., 497, 773. Panic´ O. et al. (2010) Astron. Astrophys., 519, A110. van Zadelhoff G.-J. et al. (2001) Astron. Astrophys., 377, 566. Papaloizou J. C. B. et al. (2001) Astron. Astrophys., 366, 263. Visser R. et al. (2009) Astron. Astrophys., 495, 881. Paquette C. et al. (1986) Astrophys. J. Suppl., 61, 197. Visser R. et al. (2011) Astron. Astrophys., 534, A132. Pascucci I. et al. (2009) Astrophys. J., 696, 143. Visser R. et al. (2013) Astrophys. J., 769, 19. Pascucci I. et al. (2013) ArXiv e-prints. Wada K. et al. (2009) Astrophys. J., 702, 1490. Pasek M. A. et al. (2005) Icarus, 175, 1. Walsh K. J. et al. (2011) Nature, 475, 206. Persson M. V. et al. (2013a) Astron. Astrophys., 549, L3. Wasson J. T. and Chou C.-L. (1974) Meteoritics, 9, 69. Persson M. V. et al. (2013b) Astron. Astrophys., 549, L3. Weidenschilling S. J. (1977) Mon. Not. R. Astron. Soc., 180, 57. Pilcher C. B. et al. (1972) Science, 178, 1087. Wettlaufer J. S. (2010) Astrophys. J., 719, 540. Pollack J. B. et al. (1994) Astrophys. J., 421, 615. Whipple F. L. (1950) Astrophys. J., 111, 375. Pollack J. B. et al. (1996) Icarus, 124, 62. Whittet D. C. B. (2010) Astrophys. J., 710, 1009. Pontoppidan K. M. et al. (2003) Astron. Astrophys., 408, 981. Whittet D. C. B. et al. (2007) Astrophys. J., 655, 332. Pontoppidan K. M. et al. (2004) Astron. Astrophys., 426, 925. Whittet D. C. B. et al. (2009) Astrophys. J., 695, 94. Pontoppidan K. M. et al. (2005) Astrophys. J., 622, 463. Willacy K. and Woods P. M. (2009) Astrophys. J., 703, 479. Pontoppidan K. M. et al. (2010a) Astrophys. J., 720, 887. Woitke P. et al. (2009) Astron. Astrophys., 501, 383. Pontoppidan K. M. et al. (2010b) Astrophys. J. Lett., 722, L173. Woitke P. et al. (2011) Astron. Astrophys., 534, A44. Przybilla N. et al. (2008) Astrophys. J. Lett., 688, L103. Wong M. H. et al. (2004) Icarus, 171, 153. Qi C. et al. (2004) Astrophys. J. Lett., 616, L11. Wood B. J. (1993) Earth and Planetary Science Letters, 117, 593. Qi C. et al. (2008) Astrophys. J., 681, 1396. Wood J. A. and Hashimoto A. (1993) Geochim. Cosmochim. Acta,

23 57, 2377. Wyckoff S. et al. (1991) Astrophys. J., 367, 641. Yang L. et al. (2012) p. 2023. Youdin A. N. and Shu F. H. (2002) Astrophys. J., 580, 494. Yurimoto H. and Kuramoto K. (2004) Science, 305, 1763. Zhang K. et al. (2013) Astrophys. J., 766, 82. Zhu X. K. et al. (2001) Nature, 412, 311. Zinner E. (1998) Annual Review of Earth and Planetary Sciences, 26, 147.

This 2-column preprint was prepared with the AAS LATEX macros v5.2.

24